Base Editing of TIGIT Reprograms CD155 Signaling in Natural Killer Cells to Enhance Cancer Immunotherapy Efficacy

Modifying a single base within the TIGIT gene in NK cells switches inhibitory signaling to an activating axis that enhances antitumor immunity, supporting base editing of immune cells as an immunotherapeutic strategy.

Abstract

NK cells hold great promise for cancer immunotherapy owing to their intrinsic capacity to recognize and eliminate malignant cells. Nevertheless, broad clinical deployment is hindered by NK-cell properties like poor expansion and refractoriness to genetic modification, as well as by tumor immune-evasion mechanisms. In this study, we applied base-editing technology to precisely modify signal transduction in primary human NK cells, which achieved a high editing efficiency of the T-cell immunoreceptor with Ig and ITIM domains (TIGIT) gene (>90%) in peripheral blood–derived NK (TIGIT BE-NK) cells. TIGIT editing forced tumor-derived CD155 to engage CD226 and thereby converted an inhibitory signal into an activating one that amplified NK-cell cytotoxicity. TIGIT BE-NK cells specifically targeted cancer cells across multiple tumor types and were validated as safe, with minimal off-target effects. Cryopreserved TIGIT BE-NK cells exhibited similar antitumor activity as fresh TIGIT BE-NK cells, supporting their potential as an “off-the-shelf” therapy. Combining TIGIT BE-NK cells and IL2 further improved antitumor immunity. Together, these results underscore the feasibility of using base editing to modify NK cells and significantly enhance their therapeutic potential for treating patients with cancer.

Significance:

Modifying a single base within the TIGIT gene in NK cells switches inhibitory signaling to an activating axis that enhances antitumor immunity, supporting base editing of immune cells as an immunotherapeutic strategy.

Graphical Abstract

Introduction

NK cells are promising candidates for the next generation of cancer immunotherapies because of their ability to effectively kill tumor cells (1). Unlike other immune cells, NK cells can recognize and target tumor cells independently of the MHC, making allogeneic transfer a safe option to avoid graft-versus-host diseases (2, 3). NK cells possess numerous essential attributes crucial for safer cancer immunotherapies, such as being inherently cytotoxic while devoid of the potential risks associated with cytokine release syndrome or neurotoxicity (4). The development of cancer immunotherapy based on NK cells remains challenging because of the limited lifespan and proliferative capacity of NK cells, as well as the difficulties associated with genetic modification. We have demonstrated that feeder cell–mediated ex vivo expansion of peripheral blood–derived NK (PB-NK) cells has effectively overcome their inherent limited proliferative capacity (5). The proliferative activity of PB-NK cells can be maintained for at least 5 weeks without senescence using a feeder cell–based expansion platform. Although these advances have shown promise, there remains a critical need to overcome the immune escape characteristic of cancer cells. A promising strategy involves engineering the intrinsic signaling networks in NK cells, thereby enhancing target specificity and minimizing off-target toxicity.

T-cell immunoreceptor with Ig and ITIM domains (TIGIT; also known as VSIG9, Vstm3, or WUCAM) is a receptor of the immunoglobulin superfamily. The role it plays is crucial in restricting the cytotoxicity mediated by NK cells (6, 7). TIGIT participates in a complex regulatory network involving multiple inhibitory receptors (such as CD96/TACTILE and CD112R/PVRIG), one competing activating receptor (DNAM-1/CD226), and multiple ligands [such as CD155 (PVR/NECL-5) and CD112 (nectin-2/ PVRL2); ref. 8]. Compared with resting NK cells, activated NK cells exhibit a heightened expression of TIGIT, which plays a crucial role in regulating their tumor-killing activity. However, paradoxically, they display reduced cytotoxicity against CD155-positive tumor cells compared with TIGIT-negative NK cells (9). The current study has demonstrated that the depletion of TIGIT enhances the cytotoxicity and metabolic capacity of NK cells, thereby promoting their antitumor efficacy (10). Therefore, targeting TIGIT in PB-NK cells represents a promising strategy to unleash the inherent antitumor potential of PB-NK cells.

The DNA base editor enables targeted nucleotide substitutions without inducing double-stranded DNA breaks, representing a significant advancement in the field of precision medicine. The application of base editing has been extensively employed for the disruption of target genes through the induction of pre-stop codons, interference with splicing sites, and perturbation of start codons (11, 12). However, it is crucial to consider the potential occurrence of single-guide RNA (sgRNA)–independent off-target effects (13). To address the sgRNA-independent off-target effects, we developed a base editor called cas-embedding cytosine base editor (commercially known as AccuBase), which involves the integration of deaminase into nCas9 intermediate tolerance sites (14). The AccuBase protein forms a complex with sgRNA, in which the inserted deaminase is sequestered and not in contact with any nontargeting DNA. Upon guidance by sgRNA, the AccuBase protein associates with the target DNA, inducing conformational changes that expose the deaminase domain outwardly, thereby enabling highly specific base editing while minimizing off-target effects.

Here, we present a robust and nonviral base-editing platform for primary human NK cells utilizing the AccuBase ribonucleoprotein (RNP) nucleofection technique. Our optimized protocol enables a remarkable 10,000-fold expansion of NK cells within 14 days. The expanded NK cells demonstrate exceptional purity, cytotoxicity, and viability. We have systematically fine-tuned the nucleofection conditions to achieve highly efficient knockout (KO) of the TIGIT gene in expanded PB-NK cells while ensuring that no off-target effects independent of sgRNA are observed. By modifying a single base of TIGIT, CD155/TIGIT-mediated inhibitory signal switched to CD155/CD226-mediated activation signal. In vitro and in vivo assessments confirm the potent antitumor activity and safety profile of TIGIT BE-NK cells generated through our approach. These findings strongly support the integration of base-editing technology with feeder cell–based ex vivo expansion strategies for enhancing the therapeutic potential of NK cells in clinical settings.

Materials and Methods

Cell lines and cell culture

All tumor cell lines in this study were sourced from the Shanghai Cell Bank of the Chinese Academy of Sciences and iCell Bioscience Inc. Cells were cultured in RPMI-1640 (Gibco, 61870-036) or DMEM (HyClone, SH30243.01) with 10% FBS (Nobimpex, B118-500) and 1% penicillin/streptomycin (Yeasen Biotechnology, 60162ES76). mbIL21-CD137L-K562 cells, A549 luciferase (FLuc + A549) cells (transfected with the luciferase plasmid), HT-1080 luciferase (FLuc + HT-1080) cells, and GFP-expressing H1299, A549, and HT-1080 cells (GFP + H1299, GFP + A549, and GFP + HT-1080) were established and maintained in our laboratory. All cell lines used in this study were maintained at low passage numbers (below 10). Tumor cell lines were expanded for one passage after thawing before experimental use, and low-passage reserves were regularly replenished from liquid nitrogen storage. Cell line authenticity was confirmed annually by short tandem repeat profiling, and Mycoplasma contamination was monitored quarterly using PCR-based detection. We have documented the RRIDs for all cell lines used in this study as follows: A549 (RRID: CVCL_0023), NCI-H1299 (RRID: CVCL_0060), HT-1080 (RRID: CVCL_0317), Huh7 (RRID: CVCL_0336), PLC/PRF/5 (RRID: CVCL_0485), NCI-1975 (RRID: CVCL_1511), NCI-H460 (RRID: CVCL_0459), K562 (RRID: CVCL_0004), CEM (RRID: CVCL_3496), BxPC-3 (RRID: CVCL_0186), T24 (RRID: CVCL_0554), HCT116 (RRID: CVCL_0291), HeLa (RRID: CVCL_0030), U2OS (RRID: CVCL_0042), MCF7 (RRID: CVCL_0031), MM.1S (RRID: CVCL_8792), KG-1A (RRID: CVCL_18246), HGC-27 (RRID: CVCL_1279), SKOV3 (RRID: CVCL_0532), OCI-AML-3 (RRID: CVCL_1844), NCI-H929 (RRID: CVCL_1600), and MHCC-97H (RRID: CVCL_4972).

Reagents and antibodies

Reagents were purchased as follows: 4D-Nucleofector X kit (Lonza, V4XP-3032); recombinant human IL2 (SL Pharma); D-Luciferin (IVISbrite, 122799); CellTiter-Glo Luminescent Cell Viability Assay (Promega, G7571); Stain Buffer (BD Pharmingen, 554656); and Cytolight Rapid Dye (Incucyte, 4705). The antibodies used for flow cytometry and Western blotting are indicated in Supplementary Table S1.

Flow cytometry

Data were acquired using CytoFLEX LX (Beckman Coulter, RRID: SCR_025067) and analyzed using FlowJo software (Ashland, RRID: SCR_008520). For all of the flow cytometry experiments, appropriate isotype control antibodies were used to determine the level of background staining.

Luminescent cell viability assay

The measurement of NK cell–mediated tumor cell lysis was conducted using luminescence-based assays (Promega, G7571), following previously established protocols. A total of 5 × 10³ tumor cells, NK cells, or a 1:1 mixture of both was seeded into white opaque 96-well plates and incubated for 24 hours. Subsequently, reagents were added, and the luminescent signals were quantified using the Synergy 2 Multimode Microplate Reader. The percentage of lysis was determined by employing the subsequent formula lysis % = 1 − (MIX-NK)/tumor.

Animal experiments

NOD.Cg-Prkdc^scidIl2rg^em1Smoc (M-NSG) mice (cat. #NM-NSG-001, RRID: IMSR_NM-NSG-006) were purchased from the Shanghai Model Organisms Center, and NOD/ShiLtJGpt-Prkdc^em26Cd52Il2rg^em26Cd22/Gpt (NCG) mice (strain no. T001475, RRID: IMSR_GPT:T001475) were purchased from GemPharmatech. All animal experiments were supported by the Institutional Animal Care and Use Committee at Shanghai University of Traditional Chinese Medicine. A total of 2.5 × 10⁵ FLuc + HT-1080 or 4 to 5 × 10⁵ FLuc + A549 cells per mouse were intravenously inoculated, and then mice were left untreated or treated with NK cells weekly and with or without IL2. Mouse bioluminescence or DIR fluorescence signal was detected by IVIS imaging (PerkinElmer, RRID: SCR_027425).

All animal procedures were carried out following the principles of the Basel Declaration and recommendations of Laboratory Animal Guidelines for ethical review of animal welfare (GB/T 35892-2018), Shanghai University of Traditional Chinese Medicine Institutional Animal Care and Use Committee. The protocols were approved by the Shanghai University of Traditional Chinese Medicine Institutional Animal Care and Use Committee.

Expansion of peripheral blood mononuclear cell–derived NK cells

Human peripheral blood mononuclear cells (PBMC) were obtained from the Shanghai Blood Center and frozen at 10⁷ cells/mL per cryogenic tube with 90% serum and 10% DMSO (Sigma, D2650). For frozen PBMCs, the cells were thawed 1 day before the incubation in RPMI-1640 medium (HyClone, SH30809.01) supplemented with 10% FCS, 1% penicillin/streptomycin (Yeasen Biotechnology, 60162ES76), and 200 U/mL of IL2 at 37°C in an environment of 5% CO₂. Irradiated mbIL21-CD137L-K562 cells were added at a 1:1 ratio, followed by weekly supplementation with 100 U IL2 every 3 days.

Generation of TIGIT BE-NK cells and combinatorial gene KO NK cells

NK cells were expanded from PBMCs. In this study, we chose Lonza 4D-nucleofector (RRID: SCR_023155) to generate TIGIT BE-NK cells according to the manufacturer’s protocol. Briefly, NK cells per 20 μL of mixed liquid (nucleofector:supplement 1 = 4.5:1) with base editor protein and sgRNA were seeded into 16-well strip-format nucleofection cuvettes (Lonza) and then transfected with pulse code CM189. Afterward, the cells were transferred into six-well plates and maintained by 100 U/mL IL2 and irradiated mbIL21-CD137L-K562. On days 14 to 28, TIGIT BE-NK cells were used for the experiments.

Cytosine base editor protein was purchased from KACTUS Biosystems Co. Ltd. The sgRNAs used in this study were synthesized by Nanjing GenScript Biotech Co., Ltd. and the primer sets in Supplementary Tables S2 and S3.

Deep sequencing analyses

The software Cas-OFFinder (http://www.rgenome.net/cas-offinder/) was used to predict the potential off-target sites, and 25 potential off-targets were chosen (Supplementary Table S4). The PCR products were sequenced with PE150. Burrows–Wheeler Aligner and Samtools were used to analyze the data. For whole-genome sequencing (WGS), the related DNA was extracted, and WGS was conducted at a sequencing depth of 40×. Variants were identified by GATK VariantFiltration. All kinds of changes between the nucleotides were summarized.

Real-time cell imaging

A total of 10⁴ GFP-expressing or dye-labeled tumor cells were seeded into a 96-well plate, and 6 hours later, NK cells were added. Plates were then transferred into the Incucyte S3 (Sartorius, RRID: SCR_023147), and images were captured by the Incucyte S3 system with a × 10 objective lens at 6-hour time intervals. The fluorescence intensity or cell confluence was used to quantify tumor cell growth.

Degranulation analysis

CD107a expression was detected to analyze the NK cells’ degranulation. For details, NK cells were coincubated with tumor cells in an effector-to-target (E:T) ratio of 1:1, anti-CD107a antibody or isotype was added immediately for 4 hours, and NK cells were then collected and stained with anti-CD56 antibody at 4°C for 30 minutes. The population of CD56– and CD107a double–positive cells was analyzed by flow cytometry.

Droplet Digital PCR

NCG mice were intravenously injected with FLuc + A549 cells, and then 60 tumor-bearing mice were divided into five groups (time points) with five males and five females per group. The biological samples, including whole blood, heart, liver, spleen, lung, kidney, brain, bone marrow, spinal marrow, stomach, duodenum, uterus (female), ovary (female), and testicles (male), were collected on day 1 (2 hours), day 2 (24 hours), day 4 (72 hours), day 6 (120 hours), and day 15 (336 hours) after a single intravenous injection of TIGIT BE-NK cells (1.5 × 10⁷ cells/animal). The Bio-Rad QX200 Droplet Digital PCR (ddPCR) system was used to partition each sample-PCR mixture into approximately 20,000 water-oil emulsion droplets. The reaction mixture included ddPCR Supermix for Probes (no dUTP), along with primers and a probe specifically designed to target a unique sequence within the human leukocyte antigen (HLA) alleles of TIGIT BE-NK cells (Supplementary Table S5). Following partitioning, thermal cycling was performed to amplify the target within each droplet, and the amplification was quantified based on the fluorescence emitted by the sequence-specific probes. A reaction with nontarget molecule was counted as fluorescent negative, whereas a reaction with one or more target molecules was counted positive. After amplification, data were collected using the QX200 Droplet Reader. Measured values of the TIGIT BE-NK cells should be higher than 6,000 copies/reaction.

Preparation of dye-labeled NK cells

NK cells were adjusted to a concentration of 2 × 10⁶ cells/mL. Additionally, 5 μmol/L DIR dye was added to the cell suspension, and then the cells were incubated at 37°C for 20 minutes. Following the incubation period, the cells were collected and underwent two rounds of washing to eliminate any excess dye.

Statistical analysis

The Student t test was used to test for the significance. Mean values ± SD are shown, and one-way ANOVA was used for comparisons among multiple groups. In the mouse model, survival rates were analyzed using the Kaplan–Meier method and compared with the log-rank test. P < 0.05 was considered to indicate a significant difference. Each experiment was repeated independently with similar results at least three times. The statistical analyses were performed using GraphPad Prism 9.0 software (RRID: SCR_002798) or SPSS version 26 (RRID: SCR_002865).

Results

Generation of TIGIT BE-NK cells by base editing

The mature phenotype and relative ease of collection make PBMCs a significant source of NK cells. Our methodology for generating TIGIT BE-NK cells is illustrated in Fig. 1A. Briefly, PBMCs were thawed and recovered through coculturing with 4-1BBL-mbIL21-K562 feeder cells (5). The initial expanded PB-NK cells were transfected with RNP using Lonza 4D nucleofector, and subsequently, the gene editing efficiency was evaluated using flow cytometry and DNA sequencing.

Figure 1.

Generation of TIGIT-KO NK cells by base editing. A, Schematic representation of PBMC-derived TIGIT BE-NK cell production. PBMCs were coincubated with irradiated mbIL21-CD137L-K562 cells and 100 U/mL of IL2; base editing was then performed by electrotransfection. B, The sgRNA specifically designed for base editing to knock out the TIGIT is shown in green to indicate the targeting sequence, whereas the PAM sequences are shown in red. E4, exon 4. C, The percentage and representative flow cytometry plots of TIGIT-positive NK cells on day 14 (day 14 after electroporation). Data are shown as mean ± SD of three donors. D and E, Representative flow cytometry histogram plots showing the expression patterns of activating receptors and inhibitory receptors. The antibody isotype staining results are depicted as gray peaks. D, Flow cytometry histogram plots. E, Quantitative graph of the results shown in D. Data are shown as mean ± SD of three donors. F and G, Purity (CD3⁻CD56⁺) and TIGIT expression at different time points of PB-NK and TIGIT BE-NK. Data are shown as mean ± SD of three donors. Statistical significance was determined by a two-tailed Student t test.

To enhance the efficiency of gene editing, we conducted experiments on four in vitro–transcribed (IVT) sgRNAs and ultimately selected sgRNA-2 based on its potential activity (Supplementary Fig. S1A and S1B; Fig. 1B). The subsequent step involved incubating synthetic or IVT sgRNA with AccuBase protein for transfection into NK cells. In line with the findings presented in cytokine-expanded PB-NK cells (15), synthetic sgRNA augmented gene editing efficiency (Supplementary Fig. S1C). We also assessed the potential of MaxCyte ExPERT ATx, an alternative nucleofection instrument, to achieve efficient gene editing in PB-NK cells. As expected, MaxCyte ExPERT ATx nucleofector yielded similar gene editing efficiency as the Lonza 4D nucleofector did. Furthermore, increasing the dosage of RNP resulted in enhanced gene editing efficiency (Supplementary Fig. S1D). The nucleofectors from both Lonza and MaxCyte are capable of generating TIGIT-BE cells with high quality (Supplementary Fig. S1E). The average KO efficiency, as determined from data obtained from three donors, reaches an impressive 87% (Fig. 1C). Additionally, we investigated whether base editing would alter the phenotype of NK cells. We compared the surface expression levels of activating and inhibitory receptors on PB-NK cells with those on TIGIT BE-NK cells and observed that the receptor repertoire of TIGIT BE-NK cells was nearly identical to that of PB-NK cells, except for TIGIT expression (Fig. 1D and E).

TIGIT BE-NK cells are stable and can be re-expanded after cryopreservation

Previously, we developed a feeder cell–based protocol for ex vivo expansion of NK cells (5). Here, we envisioned a streamlined protocol in which PB-NKs could be ex vivo expanded for gene editing on demand, cryopreserved for storage, thawed, and re-expanded for research and therapeutic application. Therefore, we expanded NK cells from PBMCs and transduced AccuBase RNP into PB-NKs to knock out TIGIT expression. Then TIGIT BE-NK cells were expanded and cryopreserved at day 14. After 7-day storage in liquid nitrogen, cryopreserved TIGIT BE-NK cells were thawed and re-expanded for 3 weeks (Fig. 1F). During the expansion procedure, the percentage of the TIGIT-negative population and CD56-positive population was assessed by flow cytometry once per week to investigate the stability of base editing and ex vivo expansion protocol. As shown in Fig. 1G, the percent TIGIT⁺ population of TIGIT BE-NK cells is much lower than that of PB-NK cells (28.6% vs. 74.2%) at day 8. The TIGIT⁺ population of PB-NK cells increased up to 90% at day 28, which indicated that inhibitory receptors such as TIGIT were upregulated during the activation procedure. However, the TIGIT⁺ population of TIGIT BE-NK cells decreased from 28.6% to 6.28% at day 14, which indicated that the degradation of TIGIT protein might take some time. Although NK cells were cryopreserved for 7 days, the efficiency of gene editing and purity of thawed NK cells were stable during the expansion (Fig. 1F and G). These data suggested that AccuBase RNP–mediated base editing could be inherited stably in PB-NK cells.

CD155 engagement in TIGIT BE-NK cells converts inhibitory signaling into productive activation

NK cells possess an intrinsic capacity to target solid tumor cells without chimeric antigen receptor (CAR) arming. To empower NK cells to be resistant to tumor-derived CD155 signals, we disrupted the TIGIT gene to redirect CD155 binding toward the activating receptor CD226. We cocultured tumor cells that express a high level of CD155 with PB-NK cells or TIGIT BE-NK cells. As expected, ex vivo–expanded NK cells could eliminate tumor cells effectively despite the E:T ratio being relatively low (Fig. 2A–E; Supplementary Figs. S2 and S3; Supplementary Videos S1–S5). However, the antitumor activity of PB-NK cells was impaired by TIGIT–CD155 engagement (6, 9). Percent lysis of H1299 cells of PB-NK cells was about 60% after 24-hour coculturing, whereas that of TIGIT BE-NK cells was about 80%, indicating that blockade of TIGIT–CD155 signaling by gene editing unleashes intrinsic tumoricidal activity of ex vivo–expanded NK cells. CD155 is highly expressed in a variety of solid tumors, including non–small cell lung cancer (NSCLC; ref. 16). The expression level of CD155 was assessed in multiple NSCLC cell lines and categorized into two groups based on CD155 expression: CD155^high and CD155^medium. Liquid tumor cell lines (K-562 and CEM) were employed as the CD155^low group (Fig. 3A). The killing capacity and degranulation activity of PB-NKs and TIGIT BE-NK cells were compared when cocultured with CD155^high, CD155^medium, or CD155^low tumor cells. Notably, PB-NK cells exhibited stronger cytotoxicity against H1299 cells, which have high CD155 expression, than against A549 cells that express moderate levels of CD155. This enhanced killing is likely due to the higher surface expression of the activating ligands MICA/B (17), Fas, DR4, and DR5 (18) on H1299 cells (Supplementary Fig. S4). Importantly, TIGIT BE-NK cells exhibited enhanced cytotoxicity and degranulation activity against CD155^high and CD155^medium tumor cells. However, TIGIT BE-NK cells did not exhibit any significant augmentation in the killing of CD155^low tumor cells beyond the inherent activity demonstrated by PB-NKs (Fig. 3B and C). The scope of our study was broadened through the examination of a more extensive array of cancer cell lines. Interestingly, we observed a consistent pattern in which the augmented elimination of target cells by TIGIT BE-NK cells was consistently correlated with elevated levels of CD155 expression in the tumor (Fig. 3D–F; Supplementary Fig. S5). TIGIT and CD226/DNAM-1 bind to CD155 competitively. However, the binding affinity of TIGIT and CD155 is almost two orders of magnitude higher than that of CD226 and CD155, leading to the inhibition of NK-cell cytotoxicity against CD155-positive tumor cells. To investigate whether TIGIT deficiency in NK cells would induce CD155 from tumor cells to activate CD226-mediated signaling, we subsequently employed AccuBase for simultaneous knockout of both TIGIT and CD226 in NK cells (Supplementary Fig. S6A), thereby demonstrating that TIGIT BE-NK cells elicited potent antitumor effects in CD155 wild-type tumor cells. Conversely, depletion of CD226 significantly suppressed the function of NK cells, mirroring the observed effects from double knockout of CD226/TIGIT. Moreover, depletion of CD155 in A549 and Huh7 cell lines (Supplementary Fig. S6B and S6C) resulted in cytotoxicity comparable with that induced by TIGIT BE-NK cells; similarly, this effect was reversed upon depletion of CD226 (Fig. 3G and H). Individual knockout of CD96 or PVRIG alone produced minimal effects on NK-cell effector function. Furthermore, the antitumor potency of TIGIT BE-NK cells was nearly indistinguishable from that of TIGIT/CD96/PVRIG triple–KO cells (Supplementary Fig. S6D; Fig. 3I and J). Of note, CD112 expression is substantially higher in Huh7 liver cancer cells than in A549 lung cancer cells (Supplementary Fig. S7), potentially explaining their enhanced responsiveness to PVRIG KO. TIGIT/PVR ligation prematurely terminates phosphorylation of p44/42 MAPK (ERK1/2), which resulted in the suppression of NK-cell cytotoxic function (19). In vitro, NK cells expanded with feeder cells, and IL2 exhibited robust ERK pathway activation. However, upon contact with tumor cells, ERK signaling was rapidly downregulated. TIGIT-edited NK cells sustained higher ERK activation levels, an effect that was abolished following combined knockout of TIGIT and CD226 (Fig. 3K and L). These data indicate that the knockout of TIGIT in NK cells results in the conversion of CD155-mediated inhibitory signals into activating signals.

Figure 2.

Cytotoxicity activities of TIGIT BE-NK cells in vitro. A–E, A variety of different cancer cell lines (GFP expressing or red dye labeled) were inoculated into 96-well plates in quantities of 10⁴ cells/well and left for 6 hours to allow cells to attach, then 10⁴ PB-NK or BE-NK cells were added, and images of the same field of view were recorded at different times. A, NSCLC cell line H1299 (GFP expressing). B, NSCLC cell line A549 (GFP expressing). C, Fibrosarcoma cell line HT-1080 (GFP expressing). D, Hepatocellular cell line Huh7 (red dye labeled). E, Hepatoma cell line PLC/PRF/5 (red dye labeled). Images recorded and data analyzed by IncuCyte S3, and data are shown as mean ± SD.

Figure 3.

TIGIT BE-NK cells specifically recognized target cells in a CD155-/CD226-dependent manner. A, The mean fluorescence intensity (MFI) of CD155 on H1299, H1975, A549, and H460 (NSCLC cell line), K652 (chronic myelogenous leukemia cell line), and CEM (acute lymphoblastic leukemia cell line) was assessed by flow cytometry. B, Tumor cells and PB-NK or TIGIT BE-NK cells were inoculated into 48-well plates at a 1:1 ratio, and CD107a expression on NK cells was detected by flow cytometry 4 hours later. The increase in CD107a-positive cells in the TIGIT BE-NK group was normalized to the PB-NK group. C, Tumor cells were coincubated with PB-NK or TIGIT BE-NK cells for 24 hours at an E:T ratio of 1:1, and then NK cell–mediated lysis was evaluated by luminescent cell viability assay. E:T corresponds to E:T cells. D, The mean fluorescence intensity of CD155 on U251-MG (glioblastoma cell line), Huh7 (hepatocellular cell line), HT-1080 (fibrosarcoma cell line), PLC/PRF/5 (hepatoma cell line), BxPC-3 (pancreatic adenocarcinoma cell line), T24 (urinary bladder cancer cell line), U2OS (osteosarcoma cell line), MCF7 (breast cancer cell line), HeLa (cervical carcinoma cell line), SKOV3 (ovarian cancer cell line), HCT116 (colorectal carcinoma cell line), HGC-27 (gastric cancer cell line), KG-1A, OCI-AML3 (acute myeloid leukemia cell line), and MM.1S (multiple myeloma) was assessed by flow cytometry. E, Tumor cells and PB-NK or TIGIT BE-NK cells were inoculated into 48-well plates at a 1:1 ratio, and CD107a expression on NK cells was detected by flow cytometry 4 hours later. The increase in CD107a-positive cells in the TIGIT BE-NK group was normalized to the PB-NK group. F, The correlation analysis of the result is shown in D and E. Simple linear regression analysis. G and H, The NK cell–mediated cytotoxicity was evaluated using a luminescent cell viability assay following coincubation of A549 (G) or Huh7 (H) cells with NK cells at a 1:1 ratio for 24 hours. I and J, Cytotoxic effects of individual and combined TIGIT, CD96, and PVRIG KO in NK cells. Data recorded and analyzed by IncuCyte S3; data are shown as mean ± SD. K and L, PB-NK, TIGIT BE-NK, or TIGIT/CD226 double–KO NK cells were cocultured with A549 or Huh7 tumor cells. NK cells were harvested at 0, 2, 15, and 30 minutes after stimulation, and p44/42 MAPK (ERK1/2) and phospho-p44/42 MAPK (ERK1/2; Thr202/Tyr204) activity was assessed by Western blotting. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; ns, nonsignificant.

TIGIT BE-NK cells improve tumor control in a tumor-bearing mouse model

The successful generation of TIGIT BE-NK cells enables us to evaluate the therapeutic efficacy in a solid tumor model. Using an NSCLC mouse model, we investigated whether the adoptive transfer of TIGIT-BE NK cells could boost the control of disease in tumor-bearing mice (Supplementary Fig. S8A–S8D). In brief, immune-deficient mice were intravenously injected with luciferase-expressing A549 cells (Fluc⁺A549) on day 0 and subsequently received adoptive transfer of TIGIT BE-NK cells or unmodified PB-NK cells. The burden of NSCLC was assessed using bioluminescent imaging on days 0, 7, 14, 21, 28, and 35. The mice that did not receive adoptive transfer of NK cells developed severe metastatic disease by day 35. The adoptive transfer of PB-NK cells resulted in a reduction in tumor burden in NSCLC tumor-bearing mice at day 35; however, it did not suppress the metastasis of tumor cells. Conversely, mice receiving adoptive transfer of TIGIT BE-NK cells exhibited a remarkable decrease in tumor burden and were free from developing metastatic disease.

Sarcoma is characterized by its aggressive nature and high propensity for metastasis as more than 85% of patients with sarcoma present with localized disease. Among those who develop metastatic disease, 74% present exclusively with lung metastases, whereas 9% exhibit solely bone metastases and 8% manifest both bone and lung metastases (20). We attempted to employ the adoptive transfer of TIGIT BE-NK cells in this aggressive metastatic tumor model. Similar to the NSCLC model, M-NSG mice were intravenously injected with luciferase-expressing HT-1080 cells (Fluc⁺HT-1080) on day 0, and adoptive transfer of PB-NKs or TIGIT BE-NK cells commenced on day 1. FLuc⁺ HT-1080 cells colonized the lungs within 4 hours of injection. Similar to the results observed in the NSCLC model, adoptive transfer of PB-NK cells effectively reduced tumor burden but did not exert a suppressive effect on sarcoma cell metastasis. However, mice transferred with TIGIT BE-NK cells exhibited remarkable therapeutic efficacy, resulting in tumor-free status for half of the mice (Supplementary Fig. S8E–S8H).

The cytokine IL2 exerts pleiotropic effects on the immune system and is commonly employed to sustain NK-cell viability both in vitro and in vivo (21, 22). We further confirmed the therapeutic efficacy of TIGIT BE-NK cells in combination with IL2 in the sarcoma mouse model. Combination with IL2 further augmented the antitumor efficacy of BE-NK cells (Fig. 4A–D). Notably, a low dose of BE-NK cells combined with IL2 achieved comparable efficacy to a double dose of PB-NK cells (Fig. 4E–H). The results suggest that the antitumor activity of TIGIT BE-NK cells could be enhanced when combined with a low dose of IL2.

Figure 4.

Combination treatment with TIGIT BE-NK cells and IL2 mediation improved antitumor activity. A, Scheme of the in vivo evaluation of BE-NK cells (1 × 10⁷ intravenously weekly for 2 weeks) in combination with 2 × 10⁴ U (BE-NK + IL2^L) or 4 × 10⁴ U (BE-NK + IL2^H) IL2 (intraperitoneal injection twice per week) in FLuc-expressing HT-1080 xenograft NCG mice. B, Tumor burden was determined by bioluminescent imaging. C, Luminescence was measured at days 1, 15, and 23. D, Kaplan–Meier analysis of survival in the HT-1080 xenograft model. E, Experimental design for in vivo antitumor activity assay of 1 × 10⁷ (BE-NK^L) or 2 × 10⁷ (PB-NK^H or BE-NK^H) NK cells combined with IL2 (2 × 10⁴ U intraperitoneal injection twice weekly) in FLuc-expressing HT-1080 xenograft NCG mice. F, Whole-mouse luciferase activity was measured at various time points. G, Quantification of the bioluminescence signals shown in F. H, Kaplan–Meier analysis of survival in the HT-1080 xenograft model. *, P < 0.05; ns, nonsignificant.

TIGIT BE-NK cells are available as an “off-the-shelf” therapeutic option

The key to achieving off-the-shelf availability is to maintain activity after cryopreservation (23). To validate the phenotype and function of thawed TIGIT BE-NK cells, we examined the dynamics of the TIGIT BE-NK phenotype from 0 to 24 hours after thawing by flow cytometry; rapid re-expression of the receptor for TIGIT BE-NK was observed in a short period, in which TIGIT was maintained at very low expression levels (Fig. 5A). We further validated the efficacy of cryopreserved TIGIT BE-NK cells in vivo. As shown in Fig. 5B, cryopreserved NK cells from the same donor (stored in liquid nitrogen for one week) were thawed and immediately injected alongside their fresh counterparts. Notably, cryopreserved TIGIT BE-NK cells exhibit similar antitumor activity as fresh TIGIT BE-NK cells (Fig. 5C–E).

Figure 5.

Thawed TIGIT BE-NK cells exhibit stable phenotype and cytotoxicity. A, Representative flow cytometry dot plots showing the receptors’ expression on thawed TIGIT BE-NK cells. TIGIT BE-NK cells were carefully thawed and treated with 100 U/mL IL2, and the percentage of positive cells was measured by flow cytometry at different time points. B, Schematic of the in vivo anticancer activity assay of thawed TIGIT BE-NK cells. NCG mice were inoculated with 2.5 × 10⁵ FLuc-expressing HT-1080 cells, and tumor engraftment was assessed by IVIS imaging 1 day later. On day 2 after tumor transplant, mice were left untreated or were treated with 1 × 10⁷ fresh or thawed BE-NK cells weekly; NK cells were supported by 2 × 10⁴ U IL2 via intraperitoneal injection two times weekly. C, IVIS imaging. D, Quantification of the bioluminescence signals shown in C. E, Kaplan–Meier analysis of survival in the HT-1080 xenograft model. *, P < 0.05; ns, nonsignificant.

Safety profile of TIGIT BE-NK cells: genomic stability and no detectable toxicity

Identifying potential off-target effects mediated by the AccuBase RNP complexes is critical for translating base-editing technology to the clinic. The analysis of the NK-cell transcriptome sequencing data (Supplementary Fig. S9A) revealed a significant downregulation of only TIGIT in TIGIT BE-NK cells compared with PB-NK, whereas no other genes exhibited similar changes. We further assessed the off-target effects of TIGIT editing with WGS. The software Cas-OFFinder was used to predict the potential off-target sites, and 25 potential off-targets were chosen (Supplementary Table S4). Off-targeting of approximately 15% editing efficiency was detected only in the WNT7B gene (0T22), and the off-target site was located in the intronic region, which theoretically does not affect protein expression (Supplementary Fig. S9B). Moreover, the research indicated that there was no noteworthy increase observed in the detection of single-nucleotide variations when analyzing TIGIT BE-NK in comparison with PB-NK cells (Supplementary Fig. S9C and S9D). These findings suggest that TIGIT BE-NK cells are fully validated and safe, which is promising for their potential use in clinical settings.

We further collected serum and tissue samples from healthy mice that received PB-NK or TIGIT BE-NK cell transfer at higher doses. No obvious toxicity was observed according to the results from histopathologic examination (Supplementary Fig. S9E) and hepatic and renal function detection (Supplementary Fig. S9F).

TIGIT BE-NK cells tend to localize at tumor sites in mice bearing orthotopic lung tumors

To investigate the pharmacokinetics and biodistribution of TIGIT BE-NK cells, TIGIT BE-NK cells were stained with DIR fluorescent dye and injected into tumor-bearing or nontumor mice. Major organs were harvested at two time points: day 1 (6 hours) and day 3 (72 hours) after injection. On day 1, both PB-NK and BE-NK cells showed preferential accumulation in tumor-bearing lungs compared with controls. By day 3, BE-NK cells seemed to exhibit prolonged persistence at the tumor site relative to PB-NK cells although the difference between the two groups did not reach statistical significance (Supplementary Fig. S10A and S10B). Additionally, we developed an analytic protocol based on ddPCR technology to quantify the concentration of TIGIT BE-NK cell–derived HLA gene in peripheral blood or tissue samples from NSCLC tumor-bearing mice. The concentration of TIGIT BE-NK cell–derived HLA gene in different matrices was quantitatively determined by using ddPCR at 2, 24, 72, 120, and 336 hours. The results show that TIGIT BE-NK cells are mainly distributed in the lung and peripheral blood and steadily decreased to an undetectable level within 120 hours (day 5). The highest exposure of TIGIT BE-NK cells was observed in lung samples of tumor-bearing mice, and the exposure of TIGIT BE-NK cells in blood ranks second. The exposure in the liver, kidney, heart, bone marrow, stomach, and duodenum was considerably low (Supplementary Fig. S10C; Supplementary Table S6), indicating that TIGIT BE-NK cells could infiltrate into the target organ in which the tumor cells are located.

Discussion

Recent advances in CRISPR genome-editing technology create exciting new possibilities for genetically engineered NK cells to enhance anticancer activity (15, 24, 25). However, aneuploidy and chromosomal shortening are common consequences of CRISPR/Cas9 cleavage in human T cells, highlighting potential oncogenic risks (26). Instead of causing DNA double-strand breaks, base editing directly changes a specific nucleotide base at a target site. QN-139b, a base-edited, human induced pluripotent stem cell–derived CD19/BCMA CAR NK-cell therapy developed via Cytosine Base Editor (CBE), showed promising efficacy and minimal toxicity in severe refractory systemic sclerosis (27). In this study, we developed a protocol to efficiently knock out the TIGIT gene in human primary NK cells expanded using the feeder cell–based ex vivo system. The combination of base-editing technology and the NK-cell expansion platform allows us to produce abundant highly active NK cells to meet the needs of hundreds of patients. As the base editor selected in this study yielded minimal off-target effects, TIGIT BE-NK cells are as safe as unmodified PB-NK cells (Supplementary Fig. S9A–S9F). Additionally, electroporation of AccuBase RNP avoided integration of exogenous sequence in PB-NK cells, which further ensured the stability of the phenotype and function of modified NK cells. For increasing the efficiency of gene editing, 5′ phosphate moiety removal is critical, which may prevent the recognition of exogenous RNA by innate sensors in PB-NK cells.

As a promising target in cancer immunotherapy, the number of clinical trials based on TIGIT blockade has grown quickly in recent years (28, 29). Dual PD-1 and TIGIT blockade is considered to be a promising therapeutic strategy to improve the efficacy of PD-1 blockade therapy (30). Although a synergistic effect of dual PD-1/TIGIT blockade has been observed in some preclinical animal models, anti-TIGIT drug suffered a phase III cancer setback in the clinical condition (available from Genentech). It is likely that the susceptibility of activated NK cells to fratricide driven by anti–TIGIT-activated Fc receptors may be attributed to this (10). In this study, we provide an alternative strategy to targeting TIGIT–CD155 signaling and observed convincing evidence in preclinical conditions.

Within the regulatory network of NK cells, PVR/CD155 and PVRL2/CD112 function as core ligands, interacting with the inhibitory receptors TIGIT, CD112R/PVRIG, and TACTILE/CD96, as well as the activating receptor DNAM-1/CD226. These components establish a dynamic regulatory relationship characterized by “ligand sharing, receptor competition, and signaling counterbalance,” collectively determining the cytotoxic response of NK cells against tumor cells, among which, CD226 acts as a competing receptor that binds CD155 with low affinity. The binding affinity of CD226–CD155 is almost two orders of magnitude lower than that of TIGIT–CD155 (8). Thus, tumor cell–derived CD155 binds prior to TIGIT and inhibits NK-cell antitumor response.

We individually and concurrently knocked out TIGIT, CD96, PVRIG, and CD226 in NK cells using base editing. In vitro cytotoxicity assays demonstrated that CD96 KO alone had a negligible effect on NK-cell killing, consistent with previous reports (15, 31, 32). PVRIG KO slightly enhanced NK cell–mediated killing of the hepatocellular carcinoma cell line Huh7 compared with lung cancer cells, which may be attributed to higher baseline CD112 expression in hepatocellular carcinoma (Supplementary Fig. S7). In contrast, TIGIT KO produced the most substantial enhancement of antitumor activity. Simultaneous knockout of TIGIT, CD96, and PVRIG provided only a marginal benefit over TIGIT KO alone, and this difference was not statistically significant. In vitro, feeder cell– and IL2-expanded NK cells exhibited strong activation of the ERK signaling pathways. However, upon contact with tumor cells, ERK signaling was rapidly suppressed. TIGIT BE-NK cells maintained higher levels of pathway activation—an effect that was abolished upon concurrent knockout of TIGIT and CD226. Collectively, our results establish that TIGIT serves as the dominant suppressor of peripheral blood NK-cell function within the PVR signaling network. In its absence, the ligand CD155 is redistributed to engage the activating receptor CD226, thereby potentiating NK-cell activation.

It is worth noting that the sharing of ligands between inhibitory and activating receptors is not a phenomenon observed in NK cells. For instance, NKG2A and NKG2C co-recognize HLA-E (33–35). However, within the tumor microenvironment, inhibitory signaling tends to prevail. Therefore, we propose a receptor reprogramming strategy that represents a significant advancement in the field of adoptive cell transfer therapy. By combining precise genetic modifications with sophisticated cellular engineering, we have developed a more effective and safer NK cell therapy platform. This approach not only increases the precision and reliability of NK cell–based treatments but also provides new opportunities for personalized cancer immunotherapy.

TIGIT blockade antibody therapy functions by blocking the receptor’s inhibitory signaling upon binding to NK cells and T cells. Anti-TIGIT therapies have shown success in phase II clinical trials. Tiragolumab (anti-TIGIT antibody) in combination with atezolizumab (anti–PD-L1 antibody) improved overall response rates and median progression-free survival of patients with NSCLC with PD-L1–positive tumors compared with anti–PD-1 alone (36, 37). However, this combination therapy did not elicit similar benefits in phase III clinical trials (38, 39). Therapeutic blockade of TIGIT only relieves inhibition without enhancing NK-cell proliferation, survival, or cytotoxicity. A recent study has shown that knockout of TIGIT in NK cells using CRISPR/Cas9 technology increased their antitumor response and prevented fratricide with therapeutic Fc-active TIGIT antibodies (10), which demonstrates the feasibility of TIGIT blockade antibody in combination with gene-edited NK cells for cancer immunotherapy. Our study further provides translational research results for the application of TIGIT BE-NK cells in cancer immunotherapy.

DNAM1 overexpression represents an indirect compensatory strategy that fails to eliminate the intrinsic suppression mediated by TIGIT. By merely enhancing the competing activating signal, it cannot fully counteract the persistent inhibitory checkpoint, resulting in suboptimal tumor cell killing.

In contrast to CD155-blocking antibodies, base editing–mediated TIGIT KO repurposes the “bidirectional signaling” nature of CD155 to potently activate endogenous CD226. Pan-cancer analyses of The Cancer Genome Atlas and published cohorts show that CD155 is frequently overexpressed in most solid malignancies (40), providing a biological rationale for the broad applicability of TIGIT BE-NK cells. Nevertheless, marked interpatient heterogeneity necessitates pretreatment assessment of CD155 abundance in tumor tissue to enable precision patient selection and optimize therapeutic index. The present study represents the initial report on the generation of base-edited NK cells utilizing commercially available cytosine base editor protein and synthetic sgRNA. The protocol we have developed for NK cells can also be applied to other cell types, such as T cells, facilitating the production of cell and gene therapy products. Our findings provide support for the utilization of base editing in ex vivo immune cells to augment therapeutic efficacy in clinical settings. Currently, TIGIT BE-NK cells have obtained clinical trial approval from both the National Medical Products Administration of China and the FDA of the United States for the treatment of advanced solid tumors.

Supplementary Material

Supplemental Video S2

Supplementary Video S2 captures the dynamic cytotoxic activity of PB-NK and TIGIT BE-NK cells against A549 tumor cells.(Related to figure 2B)

Supplemental Video S3

Supplementary Video S3 captures the dynamic cytotoxic activity of PB-NK and TIGIT BE-NK cells against HT-1080 tumor cells.(Related to figure 2C)

Supplemental Video S4

Supplementary Video S4 captures the dynamic cytotoxic activity of PB-NK and TIGIT BE-NK cells against Huh7 tumor cells.(Related to figure 2D)

Supplemental Video S5

Supplementary Video S5 captures the dynamic cytotoxic activity of PB-NK and TIGIT BE-NK cells against PLC/PRF/5 tumor cells.(Related to figure 2E)

Figure S1

Screening of conditions for nucleofection

Figure S2

Cytotoxicity of TIGIT-BE-NK Cells Across Cancer Cell Lines at Varying ET Ratios

Figure S3

TIGIT BE-NK Cell Cytotoxicity in Multiple Cancer Cell Lines

Figure S4

Surface expression profiles of NK Cell-regulatory molecules in H1299 and A549 cell Lines

Figure S5

Upregulation of TIGIT BE-NK degranulation levels in comparison to PB-NK

Figure S6

Generation of CD155-KO Tumor Cells and NK Cells with Combinatorial Gene Knockouts

Figure S7

CD112 Expression Levels in Various Tumor Cell Lines

Figure S8

Antitumor activity of TIGIT BE-NK cells in vivo

Figure S9

TIGIT BE-NK cells exhibited no significant off-target effects and demonstrated negligible in vivo toxicity

Figure S10

In vivo tracking of TIGIT BE-NK cells: pharmacokinetic and biodistribution analysis by IVIS and ddPCR

Table S1

Antibodies used for Flow Cytometry and Western Blot

Table S2

TIGIT-targeting sgRNA sequences

Table S3

The sgRNA sequences target CD226, CD96, PVRIG on NK cells and CD155 on tumor cells

Table S4

Primers for off-target analysis

Table S5

The primers and probe of Digital Droplet PCR

Table S6

Summary of TIGIT BE-NK Cell pharmacokinetic parameters in the blood and tissues

Supplemental Video S1

Supplementary Video S1 captures the dynamic cytotoxic activity of PB-NK and TIGIT BE-NK cells against H1299 tumor cells.(Related to figure 2A)

Data Availability

The raw sequencing data generated in this study are available in the NCBI Sequence Read Archive under the BioProject accession code PRJNA1379265. For all other original data, please contact zhushiguo@shutcm.edu.cn (S. Zhu).

Authors’ Disclosures

S. Zhu reports other support from Base Therapeutics (consulting and ownership of equity). T. Xu reports other support from Base Therapeutics (consulting and ownership of equity). W. Fan reports employment with Base Therapeutics. Liling Wang reports employment with Base Therapeutics. No disclosures were reported by the other authors.

Authors’ Contributions

C. Fang: Funding acquisition, validation, investigation, writing–original draft. G. Li: Validation, investigation, writing–original draft. M. Han: Validation, investigation. Y. Wang: Validation. W. Yu: Validation. D. Hu: Validation. J. Luo: Validation. Lixin Wang: Validation. W. Fan: Validation. Liling Wang: Validation. T. Xu: Conceptualization. X. Huang: Conceptualization, funding acquisition, investigation, methodology. C. Yao: Conceptualization, funding acquisition, validation, investigation, writing–original draft. S. Zhu: Conceptualization, resources, data curation, supervision, funding acquisition, writing–review and editing.