The dynamic interplay between tumor cells and γδT cells within the tumor microenvironment significantly influences disease progression and immunotherapy outcome. In this study, we delved into the modulation of γδT-cell activation by tumor cell ligands CD112 and CD155, which interact with the activating receptor DNAM-1 on γδT cells. Spatial and single-cell RNA sequencing, as well as spatial metabolomic analysis, from neuroblastoma revealed that the expression levels and localization of CD112 and CD155 varied across and within tumors, correlating with differentiation status, metabolic pathways, and ultimately disease prognosis and patient survival. Both in vivo tumor xenograft experiments and in vitro coculture experiments demonstrated that a high CD112/CD155 expression ratio in tumors enhanced γδT cell–mediated cytotoxicity, whereas a low ratio fostered tumor resistance. Mechanistically, CD112 sustained DNAM-1–mediated γδT-cell activation, whereas CD155 downregulated DNAM-1 expression via E3 ubiquitin ligase tripartite motif–containing 21–mediated ubiquitin proteasomal degradation. By interacting with tumor cells differentially expressing CD112 and CD155, intratumoral γδT cells exhibited varying degrees of activation and DNAM-1 expression, representing three major functional subsets. This study underscores the complexity of tumor–immune cross-talk, offering insights into how tumor heterogeneity shapes the immune landscape.

Significance: Tumor cells in different intratumoral neighborhoods display divergent patterns of ligands that regulate γδT-cell activation, highlighting multilevel regulation of antitumor immunity resulting from the heterogeneity of intercellular interactions in the tumor microenvironment.

Neuroblastoma is the most common and fatal pediatric tumor, accounting for 15% of all cancer-related pediatric deaths worldwide (1). Recent data reveal that it is the second most common cancer among children ages 1 to 4 years in China (2). Neuroblastoma exhibits a heterogeneous clinical course, ranging from spontaneous resolving to devastating progression despite extensive multimodal treatments. Patients classified as low or intermediate risk have an overall survival rate of greater than 95%, whereas less than 50% of high-risk patients outlast tumor recurrence and metastasis (3).

The limited therapeutic efficacy of traditional chemoradiotherapy and advanced immunotherapies, such as anti-GD2 antibody and GD2 chimeric antigen receptor T cells, is primarily attributed to the strong immunosuppressive tumor microenvironment (TME) and the intrinsic resilience of heterogeneous tumor cells (4). The cross-talk between tumor and immune cells is complex and dynamically regulated within the TME. Tumor cells can upregulate membrane proteins, including PD-L1, FGL1, and B7H4, upon cell-to-cell contact, activating immune checkpoint pathways that inhibit immune responses and enhance tumor survival (57). Through receptor–ligand interactions, tumor cells disarm immune cells by downregulating activating receptors (8). Intertumor and intratumor heterogeneity, encompassing cytogenetic diversity, metabolite composition, proliferative activity, and stemness status, further complicates immune–tumor cell interactions (9, 10). Within a single tumor, spatial heterogeneity influences immune cell infiltration, TME remodeling, tumor cell metastasis, and patient survival outcomes (11).

DNAM-1, an activating receptor expressed by γδT cells, CD8 T cells, and NK cells, is crucial for their cytotoxic activity against tumor cells (12). Reduced DNAM-1 expression in these immune cells is linked to impaired tumoricidal function in patients with neuroblastoma, melanoma, acute myeloid leukemia, and T-cell lymphoma (1317). DNAM-1 recognizes the ligands CD112 and CD155 on tumor cells, competing with inhibitory receptors such as TIGIT, CD96, and CD112R (18, 19). The DNAM-1–ligand interaction is essential for targeting various tumor cells, including those in myeloma, neuroblastoma, and melanoma (2022). Reduced CD112 expression in tumors impairs NK cell recognition and cytotoxicity, leading to immune evasion and poor prognosis (23). CD155 also mediates optimal killing of tumor cells with engagement of DNAM-1 (24). However, recent studies have reported that CD155 expression on tumors, through interacting with DNAM-1, inhibits CD8 T-cell and NK-cell function and lack of CD155 expression enhances tumor susceptibility to immune cell killing (25, 26). As functional molecules binding to activating receptors on immune cells, multiple ligands including CD112 and CD155 express in a remarkable variance across tumor cells. Different compositions and redundancies of these ligands synergistically work for immune surveillance and tumor escape (27). Under certain pathophysiologic conditions, such as viral infection, both CD112 and CD155 may be downregulated simultaneously (28, 29). Modulation of one ligand can influence the other; for instance, upregulation of CD112 in tumor cells can be accompanied by increased DNAM-1 and decreased TIGIT expression in CD8 T cells in CD155-deficient mice, suggesting a compensatory mechanism for immune surveillance against methylcholanthrene-induced tumor (30). However, the regulation of CD112 and CD155 on tumor cells and their interaction with DNAM-1 on immune cells in the context of immune surveillance or tumor escape remains incompletely understood.

Among immune cells, γδT cells are critical players in cancer immunotherapy because of their MHC-independent nature and high tumor infiltration. Our previous studies have shown that patients with neuroblastoma exhibited increased numbers of γδT cells, which are characterized by impaired function, decreased DNAM-1 expression, and enhanced Th17 polarization (13, 14). Within tumor, various factors influence γδT-cell infiltration, activation, and functional plasticity (31, 32). However, the association between tumor cell heterogeneity and the functional status of γδT cells remains unclear, particularly with regard to how intertumoral and intratumoral variations in CD112 and CD155 expression affect DNAM-1–mediated γδT-cell activation. These insights are crucial for understanding how ligand heterogeneity influences immune cell function and disease outcomes.

In this study, using bulk RNA sequencing (RNA-seq), spatial analysis, and single-cell RNA-seq (scRNA-seq), we identified intertumoral and intratumoral heterogeneity in γδT cells, with varying degrees of functional activation and DNAM-1 expression that closely associate with patient prognosis and survival. This functional heterogeneity of γδT cells results from interactions with tumor cells, displaying differential expression levels and spatial distributions of CD112 and CD155. scRNA-seq and spatial metabolomic analyses revealed that tumors with distinct CD112/CD155 expression ratios belonged to different subsets characterized by their differentiation status, metabolic rate, and risk level. Functional studies demonstrated that tumors with a high CD112/CD155 ratio were susceptible to γδT-cell cytotoxicity, whereas tumors with a low ratio exhibited resistance to γδT-cell killing. As a result of DNAM-1–ligand interactions in the context of tumor heterogeneity, three major γδT-cell functional subsets emerged. IFNγ+ type 1 γδT cells, which express higher DNAM-1, were abundant in tumors with high CD112/CD155 ratios. In contrast, two subsets of γδT cells, the IL17+ type 3 subset and a novel IFNγ IL17 CXCR6hi subset, which exhibit markedly low DNAM-1 expression, were enriched in tumors with low CD112/CD155 ratios. This study highlights the multilevel suppression of γδT-cell activation as a consequence of heterogeneous activating receptor–ligand interactions within the TME.

Human subjects and sample collection

A total of 91 patients [45 boys and 46 girls; mean age, 47.1 ± 4.7 months; International Neuroblastoma Staging System (INSS) stage: stage I/II/III, 53.8%; stage IV, 29.7% (16.5% patients without stage data); Children’s Oncology Group risk: low/intermediate risk, 54.9%; high risk, 29.7% (15.4% patients without risk data); MYCN amplification, 16.5%; no MYCN amplification, 65.9% (17.6% patients without MYCN data)] with neuroblastoma and 53 healthy children (30 boys and 23 girls; mean age, 50.3 ± 5.2 months) under regular physical examination were enrolled between January 2020 and February 2023 from Beijing Children’s Hospital. The study was approved by the Medical Ethics Committee of Beijing Children’s Hospital, Capital Medical University, which acts in accordance with recognized ethical guidelines defined by the Declaration of Helsinki. Written informed consent for research purposes was provided by all participants and their parents or legal guardians. Peripheral blood (PB) or bone marrow (BM) samples were collected in BD Vacutainer plastic blood collection tubes with EDTA K2 as anticoagulant. Serum samples were collected in tubes without anticoagulant by a centrifugation at 600 g for 5 minutes, and the aqueous phase was taken for analysis. Neuroblastoma tumor tissues were collected in saline solution and subjected to subsequent experiments within 1 hour or stored in −80°C. The clinical characteristics and the follow-up information are summarized in Supplementary Table S1.

Cell lines

Neuroblastoma cell lines SH-SY5Y (RRID: CVCL_0019), SK-N-BE2 (RRID: CVCL_0528), and IMR-32 (RRID: CVCL_0346) were purchased from ATCC. CHLA-255 cells (RRID: CVCL_AQ27) were kindly provided by Prof. Shahab Asgharzadeh from Children’s Hospital Los Angeles, Los Angeles, CA. SH-SY5Y and SK-N-BE2 cell lines were cultured in DMEM (Corning) supplemented with 10% FBS (Gibco, Invitrogen) and 1% penicillin/streptomycin, IMR-32 cultured in minimum essential medium (Corning), and CHLA-255 cultured in Iscove’s DMEM (IMEM; Corning). Cells were cultured in a humidified cell incubator at 37°C and with 5% CO2.

Neuroblastoma mouse tumor model

Five- to six-week-old female nude mice (CAnN.Cg-Foxn1nu/Crl, RRID: IMSR_CRL:194) were purchased from Vital River Laboratories. The mice were bred and maintained under specific pathogen-free conditions with sterilized food and water and housed in a barrier facility under a 12-hour light/dark cycle. The mice were randomly divided into two groups and injected with SK-N-BE2-KoCD155 cells or SK-N-BE2-NC cells (n = 6). In total, 5 × 106 cells in 200 μL PBS containing Matrigel (Corning) were injected subcutaneously into the right flank of each mouse. After 34 days, tumors were measured, and mice in each group were randomly divided into two subgroups before receiving γδT-cell adoptive transfer (SK-N-BE2-KoCD155, SK-N-BE2-KoCD155 + γδT, SK-N-BE2-NC, and SK-N-BE2-NC + γδT; n = 3 each). In γδT-injection groups, 1 × 107 γδT cells were injected through the vena caudalis every 4 days for five times. All tumors were measured every 4 days based on which the growth curves were generated. By the end of the experiment, mice were sacrificed, and tumors were excised and weighed. The tumor sections were then subjected to IHC for CD3 and T-cell receptor γδ (TCRγδ) staining using CD3 polyclonal antibody (Proteintech, RRID: AB_1939430) and TCRγδ antibody (Novus Biologicals, RRID: AB_2861306).

For the neuroblastoma orthotopic tumor model, 5- to 6-week-old female NSIG mice (NOD-PrkdcscidIL2rgtm1 mice, HFKbio) were used. The mice were randomly divided into two groups, and SK-N-BE2-luc-KoCD155 cells or SK-N-BE2-luc-NC cells (2 × 106 cells in 35 μL of PBS containing Matrigel; n = 6) were seeded into the left kidney. Ten days after, tumor growth was monitored in vivo for fluorescence intensity using the IVIS Spectrum Imaging System (PerkinElmer, RRID: SCR_018621) 12 minutes after injection of D-luciferin to the abdominal cavity. Mice were then randomly selected for γδT-cell immunotherapy or as control including groups SK-N-BE2-luc-KoCD155, SK-N-BE2-luc-KoCD155-γδT, SK-N-BE2-luc-NC, and SK-N-BE2-luc-NC-γδT (n = 3). In γδT-injection groups, 1 × 107 γδT cells were injected through the vena caudalis every 4 days for three times. The radiance of each tumor was calculated and analyzed statistically.

Bioinformatic analysis

The expression patterns of CD112, CD155, and DNAM-1 in neuroblastoma (INSS stage, high risk, MYCN amplification, or patient death) were analyzed using GSE49711 (33) and GSE25624 (34) datasets from the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/). These data were originally generated from Illumina HiSeq 2000 (Homo sapiens; Illumina). Datasets were quality-controlled before differential genetic screening analysis. A gene expression matrix with gene-level expression was quantified using Cufflinks on RefSeq genes and reported as log2 [1 + FPKM (fragments per kilobase of exon model per million mapped fragments)]. At the beginning, we performed background correction and normalization to verify conformance of parallel trials using language R software (R version 4.0.4, RRID: SCR_001905; ref. 35). Then we performed differently expressed gene (DEG) analysis. Both P value <0.05 and |log2 fold change| >1 were considered the critical values for DEG screening based on the paired t test of the “limma” R package (36).

PB mononuclear cell and BM cell isolation

Freshly isolated EDTA anticoagulated blood or BM was diluted with PBS solution and layered carefully on Ficoll-Hypaque density gradients. After centrifuged at 1,000 g for 20 minutes at room temperature, the interphase cell layer was carefully transferred into a 15-mL tube. The 15-mL tube was then filled with 10 mL PBS and centrifuged at 600 g for 5 minutes. The cell pellet was resuspended, and the viability of isolated PB mononuclear cells (PBMC) or BM cells was determined by trypan blue exclusion staining.

γδT-cell culture

PBMCs were cultured in 10% FBS RPMI 1640 medium (Gibco, Invitrogen) with 10 µg/mL of pamidronate (PAM; Sigma-Aldrich) in a humidified cell incubator at 37°C and with 5% CO2. Recombinant human (rh) IL2 (R&D Systems) was added every 3 days to a final concentration of 25 ng/mL with fresh culture medium being replenished. After a 14-day culture, the purity of γδT cells was determined by flow cytometry with PE-labeled anti-TCRγδ antibody (BioLegend). γδT-cell expansion was considered as successful when the TCRγδ+ population represents at least 85% of total cells in the culture.

Tumor dissociation

Fresh neuroblastoma tumor tissues were collected and dissociated using Tumor Dissociation Kit, human (Miltenyi Biotec) according to the manufacturer’s instructions. Briefly, fat, fibrous, and necrotic areas were removed from the tumor sample, and tumors were cut into 2- to 4-mm small pieces. The tissue pieces were mixed with enzymes H, R and A and incubated for 30 minutes at 37°C under continuous rotation. The cell suspension was washed three times using sterile PBS buffer after filtering it through a 70-μm cell strainer.

Concentration measurement of serum CD112 and CD155

Blood serum samples were collected from patients with neuroblastoma and healthy children. CD112 and CD155 concentrations were measured by ELISA kits (Novus Biologicals and Genway, respectively) according to the manufacturer’s instructions.

Carboxyfluorescein succinimidyl ester proliferation assay

After 4-day culture, γδ cells were resuspended in 2 mL RPMI 1640 medium and incubated with 2 µmol/L carboxyfluorescein succinimidyl ester (BioLegend) at 37°C for 10 minutes. The reaction was terminated by cold medium with 10% FBS and placed on ice for 5 minutes. The stained γδT cells were washed and resuspended with RPMI 1640 medium plus 10% FBS. γδT cells were then seeded (1 × 106/mL) to a 96-well plate in the presence of 25 ng/mL IL2 and 200 ng/mL rhCD112/rhCD155 protein (Abcam) and cultured for another 4 days. Cells were analyzed using the FACSFortessa flow cytometer (BD Biosciences). Cell division reflected by carboxyfluorescein succinimidyl ester fluorescence dilution was calculated and represented using FlowJo software (BD Biosciences, RRID: SCR_008520).

Flow cytometry

Freshly isolated PBMCs, tumor tissue cells, or cultured γδT cells were resuspended in cell staining buffer (PBS + 5% FBS). For surface marker labeling, 1 × 106 cells were incubated with different panels of fluorochrome-conjugated antibodies. For cytokine detection, cells were treated with 50 ng/mL PMA, 1 μg/mL ionomycin (Beyotime Biotechnology), and GolgiStop protein transport inhibitor (BD Biosciences) for 5 hours. Intracellular staining was done with different cytokine antibodies following cell permeabilization with Fixation/Permeabilization Solution Kit (BD Biosciences) according to the manufacturer’s protocol. Isotype antibodies conjugated with same fluorochrome were used to exclude the nonspecific staining. All antibodies were obtained from BioLegend and previously titrated to determine the best working concentration. After incubation for 30 minutes at 4°C in dark (45 minutes for intracellular staining), samples were washed in PBS before finally resuspended in 500 μL PBS and subjected to flow cytometry. The fluorochrome-conjugated antibodies are shown in the following paragraph. Cell events were acquired using a BD FACSCalibur or a FACSFortessa flow cytometer (BD Biosciences). Data were analyzed using FlowJo software (BD Biosciences). The following antibodies were used for flow cytometry:

FITC/PE/APC/PE-Cy7-TCRγδ (RRID: AB_1575108/AB_1089218/AB_1089214/AB_2562891), APC/APC-Cy7-TCRαβ (RRID: AB_10612569/AB_2566601), FITC/BV786-DNAM-1 (RRID: AB_2228763/AB_2721560), APC/PE-Cy7-TIGIT (RRID: AB_2632732/AB_2632928), BV421-CD96 (RRID: AB_2629537), FITC-PD-1 (RRID: AB_940479), PE/PE-Cy7-TIM3 (RRID: AB_2116576/AB_2561720), FITC/BV421-CD45 (RRID: AB_2566368/AB_2687375), PE/BV711/PE-Cy7-CD3 (RRID: AB_571913/AB_2562907/AB_314052), PE-CD4 (RRID: AB_1937246), APC-CD8 (RRID: AB_314116), PE-CD19 (RRID: AB_314238), APC-CD68 (RRID: AB_2275735), PE-CD16 (RRID: AB_314208), APC-CD56 (RRID: AB_2563913), PE-Cy7-CD27 (RRID: AB_2561919), BV510-CD45RA (RRID: AB_2561947), FITC/PE-Cy7-Granzyme B (RRID: AB_2687030/AB_2728381), PE-Perforin (RRID: AB_314704), APC-IFNγ (RRID: AB_315443), PE-CD112 (RRID: AB_2269088), APC-CD155 (RRID: AB_2565815), and APC-Annexin V.

Cytotoxicity assays

γδT cell–mediated cytotoxicity against neuroblastoma cell lines was assessed using a CytoTox 96 Non-Radioactive Cytotoxicity Assay kit (Promega Corporation) according to the manufacturer’s protocol. Briefly, before coculture with the target tumor cell line, γδT cells from PBMCs of healthy children were in vitro expanded by PAM stimulation in the presence of recombinant cytokines in a round-bottom 96-well culture plate. Wells seeded only with γδT cells (4 × 105/well) served as the spontaneous lactate dehydrogenase (LDH) release control for effectors, and wells seeded only with neuroblastoma cells (2 × 104/well) served as the spontaneous LDH release control for target cells. The coculture wells were seeded with γδT cells and neuroblastoma cells at a ratio of 20:1. Cells were centrifuged at 250 g for 4 minutes at 20°C after incubation at 37°C for 7 hours. The lysis solution was added to the target cell control wells 45 minutes prior to supernatant harvest for determination of maximal LDH release. Then a total of 50 µL supernatant from each well was transferred to a flat-bottom 96-well plate preloaded with 50 µL/well reconstituted substrate mix. Following incubation at room temperature in dark for 30 minutes, 50 µL stop solution was added, and absorbance at 490 nm was read using the TriStar2 LB 942 Multimode Reader (Berthold). The cytotoxicity of effector γδT cells to target tumor cells was calculated as [(experimental − effector spontaneous − target spontaneous)/(target maximal − target spontaneous)] × 100. In some experiments, γδT cells used in cytotoxicity assays were pretreated with 1,000 ng/mL rhCD112 or rhCD155 protein (precoated on plates overnight) for 24 hours or neutralizing antibodies against CD112 (10 μg/mL, R&D Systems) or CD155 (10 μg/mL, GeneTex, RRID: AB_809606).

Real-time cellular analysis assays

SK-N-BE2 cells were seeded into 96-well E-plates (Agilent Technologies) at 2 × 104 cells per well and monitored overnight using the xCELLigence real-time cell analyzer (Agilent Technologies, RRID: SCR_019571) according to the manufacturer’s instruction. xCELLigence real-time cellular analysis (RTCA) was used to measure cellular adhesion through electrical impedance, which is converted to an cell index. Briefly, when the cell index reached a plateau, γδT cells were added to 96-well E-plates in triplicate at an E/T ratio 10:1. The cells in the E-plates were monitored for another 16 hours using the RTCA system, and impedances (normalized to the time of γδT-cell addition) were plotted over time.

Transfection and infection

The expression vectors pcDNA3.1-CD112-His was constructed. The pcDNA3.1-CD112-His plasmid was transfected into IMR-32 cells using Lipofectamine 3000 (Invitrogen) according to the manufacturer’s instructions. The cell lysate was harvested 48 hours after transfection for subsequent experiments (IMR-32-vector vs. IMR-32-CD112). Lentiviral particles (LV3-shCD112-GFP-puromycin) containing the shRNA sequence 5′-GCA​ACT​ACA​CTT​GCG​AGT​TTG-3′ were prepared for CD112 knockdown. CHLA-255 cells were infected with the LV3-shCD112-GFP-puromycin lentiviral particles, and cells stably infected with the expression vector containing GFP were acquired through 1 μg/mL puromycin (Beyotime Biotechnology) selection for 7 days. CHLA-255 cells with CD112 stably knocked down were designated as CHLA-255-shCD112, and the control cells were named as CHLA-255-NC.

Lentiviral particles containing CD155 guide RNA-seq and Cas9 gene (LV-U6-CD155sgRNA-EF1a-Cas9-FLAG-CMV-EGFP-P2A-puromycin, LV-CD155-sgRNA: 5′-CGG​GAT​GCC​CAA​TAC​GAG​CC-3′) were prepared for knockout CD155 expression. SK-N-BE2 cells or SK-N-BE2-luc cells were infected with lentiviral particles, and cells stably infected with the expression vector containing GFP were acquired through 1.5 μg/mL puromycin selection for 7 days. The SK-N-BE2 cells with CD155 stably knocked out were sorted using the BD FACSAria II flow cytometer (BD Biosciences) and designated as SK-N-BE2-KoCD155 or SK-N-BE2-luc-KoCD155 (hybrid cell clones). Their respective control cells were designated as SK-N-BE2-NC or SK-N-BE2-luc-NC.

The Tet-on controllable expression system was achieved by a transfection with both lentiviral particles for CD155 overexpression (TRE3GS-CD155-GFP-Ubi-TetR-IRES-neomycin) and CD112 knockdown (TRE3GS-TurboRFP-shCD112-Ubi-TetR-IRES-puromycin) and subjected to 1.3 mg/mL G418 (TransGen Biotech) selection for 14 days following 1.5 μg/mL puromycin selection for another 7 days. These cells were designated as SK-N-BE2-CD155-shCD112 cells and maintained in complete DMEM containing different concentrations of doxycycline (0 nmol/L, 2 nmol/L, 20 nmol/L, 200 nmol/L, 2 µmol/L, and 20 µmol/L; Solarbio Life Sciences). The expression of CD112 and CD155 in these cells was detected by qPCR and Western blot analyses.

When E3 ubiquitin ligase tripartite motif–containing 21 (TRIM21) knockdown was required, the Neon Transfection System (100 µL Buffer T, Invitrogen) was used for siRNA transfections into γδT cells. Briefly, expanded γδT cells were activated using antibodies against CD3 (coated, 5 µg/mL) and CD28 (soluble, 5 µg/mL) for 48 hours before transfected using 200 pmol of siRNA for every 2 × 106 cells using the following settings (37): 1,400 V, 10 ms, and 3 pulse. Cells were then dispensed into 2 mL prewarmed culture medium in a 12-well plate for 48-hour culture and subjected to FACS, Western blot analysis, and qPCR analysis. The sequence of siRNA was siRNA-TRIM21, 5′-GGA​AGU​CAC​UUC​ACC​AUC​Att-3′.

Coculture of γδT cells and neuroblastoma cells

γδT cells were cocultured with different types of neuroblastoma cells, including dissociated primary neuroblastoma tumor cells; wild-type SH-SY5Y, SK-N-BE2, CHLA-255 and IMR-32 cells; and plasmid-transfected or lentivirus-infected neuroblastoma cell lines. In a 24-well plate, 5 × 105 γδT cells were cocultured with 5 × 106 primary neuroblastoma cells or 2.5 × 106 cell lines for 24 hours. In some coculture experiments in which Transwell chambers were applied to prevent direct cell contact, different ratios of γδT cells to SH-SY5Y cells (1:0.25, 1:0.5, 1:1, 1:5, and 1:10) were set. Neutralizing antibodies against CD112 (R&D Systems) or CD155 (GeneTex; 10 μg/mL) were used in some of the coculture assays before expression of DNAM-1, TIM3, and PD-1 being measured. The expression of DNAM-1 was evaluated 6 to 8 hours after some cocultures treated with proteasome inhibitor MG132 (1 µmol/L), lysosome inhibitor NH4Cl (20 mmol/L), or Src kinase inhibitor PP2 (5 µmol/L; Sigma-Aldrich).

qPCR

Total RNA from neuroblastoma tumor tissues, neuroblastoma cell lines, or γδT cells was isolated using TRIzol reagent (Invitrogen). Reverse transcription was performed according to the standard protocols using a RevertAid II First Strand cDNA Synthesis Kit (Thermo Fisher Scientific Inc.). qPCR was performed as previously described (14). GAPDH was amplified as an internal standard. The following primer sequences were used:

  • CD112-F, 5′-GCC​GCC​ATC​ATT​GCT​ACT​GC-3′;

  • CD112-R, 5′-TTC​GCT​TTT​GGG​GTC​GGT​G-3′;

  • CD155-F, 5′-CAG​GTG​GAC​GGC​AAG​AAT​GTG-3′;

  • CD155-R, 5′-GGC​CTC​ATT​CTG​GCC​AAG​GTA-3′;

  • DNAM-1-F, 5′-AAA​AGA​TCC​AGC​CCC​GTC​AG-3′;

  • DNAM-1-R, 5′-GCA​AGT​AGC​AGC​GGT​AAA​GCC-3′;

  • TRIM21-F, 5′-CCA​CAG​CTT​CTG​CCA​GGA​AT-3′;

  • TRIM21-R, 5′-TTG​GCT​AGC​TGT​CGA​TTG​GG-3′.

Cell proliferation assay

Cell proliferation was measured using a Cell Counting Kit-8 detection kit (Dojindo Molecular Technologies). SK-N-BE2 or CHLA-255 cells were seeded at a concentration of 3,000 or 6,000 cells per well in a 96-well plate and cultured in either complete DMEM or IMEM. At the indicated time points, 10 μL of Cell Counting Kit-8 solution was added to each well, and the plate was then incubated at 37°C for 120 minutes. The absorbance value of each well was read at 450 nm.

Wound-healing assay

Wound-healing assay was performed to determine the migration rate of tumor cells. SK-N-BE2 or CHLA-255 cells were cultured in a six-well plate until 80% confluence, and in the middle of the cell monolayer, a wound was carefully scrapped using a sterilized pipette tip. Then the medium was replaced with complete DMEM or IMEM. Photomicrographs were taken immediately or 72 hours after scrapping. The wound widths in the microscopic images were measured at different time points using ImageJ software (NIH, RRID: SCR_003070). The percentage of wound healing was calculated based on the initial wound width at 0 hour.

Cell migration assay

Cell migration was analyzed using Transwell chambers (8 μm pore size, Corning). Cells were cultured in serum-free DMEM or IMEM for 12 hours before being seeded into the upper chamber at a density of 1 × 105 cells per well in 250 μL serum-free DMEM or IMEM. A measure of 500 μL complete DMEM or IMEM was added to the lower chamber. The chambers were disassembled after incubation at 37°C with 5% CO2 for 48 hours. The membranes were then fixed with 4% paraformaldehyde/PBS and stained with 2% crystal violet for 10 minutes. The absolute cell numbers were counted from images captured using a microscope (100× magnification; IX73, Olympus).

Coimmunoprecipitation

Cells were lysed in RIPA buffer (Sigma-Aldrich) containing 1% protease inhibitor cocktail (Roche). Pierce Co-Immunoprecipitation Kit (Thermo Fisher Scientific Inc.) was used according to the manufacturer’s instructions. Briefly, total protein extract was incubated with the control agarose resin at 4°C for 60 minutes for preclear. Anti–DNAM-1 antibody or control rabbit IgG (10 μg) was immobilized on coupling resin by incubation on a rotator at room temperature for 120 minutes. Total protein extract was further incubated with coupling resin overnight at 4°C under agitation. Coupling resin conjugated with immune-captured samples was washed five times with Modified Dulbecco’s PBS. Samples were eluted with 60 μL elution buffer and subjected to SDS-PAGE or Western blot analysis.

Western blot analysis

The PAM-expanded γδT cells or neuroblastoma cell lines or tumor tissues were collected and lysed in RIPA buffer (Sigma-Aldrich) containing 1% protease inhibitor cocktail and 1% PhosSTOP (Roche). The protein concentrations were determined through bicinchoninic acid (BCA) protein assays (Pierce), and the whole-cell lysate was fractionated by 12.5% SDS-PAGE and electro-transferred onto polyvinylidene difluoride membranes (Hybond, GE Healthcare). After blocking the nonspecific binding sites with 5% milk in TBS containing 0.1% Tween 20 (TBS-T) for 1 hour at room temperature, the membrane was incubated with the primary antibodies overnight at 4°C. They were then extensively washed with TBS-T following a 1-hour incubation at room temperature with horseradish peroxidase–conjugated anti–mouse IgG or anti–rabbit IgG antibodies (1:5,000; Cell Signaling Technology). After three additional washes in TBS-T, the signal was visualized with ECL substrate (Applygen) and imaged using FUSION Solo S (Vilber Lourmat).

Western blot analysis was performed using the primary antibodies against tAkt (RRID: AB_2225340); pAkt (Ser473; RRID: AB_2315049); tERK1/2 (RRID: AB_390779); pERK1/2 (Thr202/Tyr204; RRID: AB_2315112; Cell Signaling Technology); β-actin (RRID: AB_2923704; Proteintech); DNAM-1 (Santa Cruz Biotechnology, RRID: AB_3096147; Abcam, RRID: AB_2049406); CD112 (Solarbio Life Sciences) and CD155 (Novus Biologicals); and TRIM21 (ABclonal, RRID: AB_2763983).

Calcium flux detection

γδT cells were pretreated with 1,000 ng/mL CD112 or CD155 protein (precoated on plates overnight) for 24 hours. Being incubated with 10 µmol/L Fluo 4-AM (Dojindo Molecular Technologies) in RPMI 1640 for 1 hour, γδT cells were washed twice with PBS and stimulated by 5 ng/mL PMA and 10 ng/mL ionomycin. Calcium flux changes were detected using a BD FACSFortessa flow cytometer (BD Biosciences). Data were acquired continuously, and the first 60 seconds were used as the basal calcium flux line before any stimulant was added. Calcium flux was recorded for 5 minutes before the second stimulation. The calcium flux results were used to benchmark the levels of activation.

Immunofluorescence confocal microscopy

SK-N-BE2 cells were stained using PKH67 Green Fluorescent Cell Linker Kit (Sigma-Aldrich). γδT cells were cocultured with SK-N-BE2 cells for 4 hours before being transferred to Petri dishes pretreated with poly-D-lysine (Beyotime Biotechnology). After 30-minute incubation, cells were fixed with 4% paraformaldehyde/PBS for 10 minutes, blocked by 3% BSA/PBS for 30 minutes, and incubated with primary antibodies against DNAM-1 (Abcam, RRID: AB_726268) at 4°C overnight. After three washes in PBS with 0.05% Tween 20, cells were incubated for 1 hour at room temperature with Alexa Fluor 555–labeled anti–mouse IgG (Beyotime Biotechnology, RRID: AB_2890132) followed by three additional washes in PBS with 0.05% Tween 20. Cell-to-cell interaction and DNAM-1 expression were recorded using a Leica TCS SP8 confocal microscope (Leica Microsystems Inc.).

For the detection of DNAM-1 and TRIM21 colocalization, γδT cells were treated with rhCD155 for 24 hours before 0.3% Triton X-100/PBS was added for cell permeabilization following fixation. Primary antibodies against TRIM21 (ABclonal, RRID: AB_2763983) and Alexa Fluor 488–labeled secondary anti–rabbit IgG were used (Abcam, RRID: AB_2630356).

Library preparation and bulk RNA-seq

Quantity and integrity of RNA were measured with the RNA 6000 Nano Assay Kit employing a Bioanalyzer 2100 system (Agilent Technologies, RRID: SCR_019715). Total RNA was transcribed into double-strand cDNA. After adenylation of 3′-ends of DNA fragments, adapters with a hairpin loop structure were ligated to prepare for hybridization. The DNA fragments preferentially 370 to 420 bp in length were purified using the AMPure XP system (Beckman Coulter). DNA fragments were multiplied by PCR and purified with AMPure XP beads for the complete library construction. The libraries were then sequenced using Illumina NovaSeq 6000 (Illumina, RRID: SCR_016387).

Data analysis for RNA-seq

The image data measured using the high-throughput sequencer were converted into sequence data (reads) through CASSAVA base recognition. Raw data (raw reads) of FASTQ format were first processed through fastp software (RRID: SCR_016962). Reference genome and gene model annotation files were downloaded from the genome website directly. The index of the reference genome was built, and paired-end clean reads were aligned to the reference genome using HISAT2 (v2.0.5, RRID: SCR_015530). featureCounts (v1.5.0-p3) was used for counting the read numbers mapped to each gene. FPKM was calculated based on the length of the gene and read count mapped to this gene for the possibility of reading errors. Prior to differential gene expression analysis, the read counts were adjusted through one scaling normalized factor. Differential expression analysis of two groups was performed using R software (R version 4.0.4) with P value <0.05 and |log2 fold change| >0 as the critical values for DEGs. Afterward, Gene Ontology including biological process, cell composition, and molecular function, and Kyoto Encyclopedia of Genes and Genomes (KEGG) were generated using the functional enrichment analysis tool “clusterProfiler” (38). Finally, hierarchical clustering was performed to display the distinguishable gene expression pattern among groups. Gene Set Enrichment Analysis (GSEA) is a computational approach to determine if a predefined gene set exhibits consistent difference between two biological states (39). We used the local version of the GSEA analysis tool (http://www.broadinstitute.org/gsea/index.jsp) with KEGG datasets as input for GSEA. Normalized enrichment score and nominal P value were calculated, and the top 50 GSEA pathways were shown according to the P value.

Single-cell RNA library construction and sequencing

DNBelab C Series High-throughput Single-Cell RNA Library (MGI) was utilized for scRNA-seq library preparation. In brief, the single-cell suspension was converted to a barcoded scRNA-seq library through steps including droplet encapsulation, emulsion breakage, mRNA-captured bead collection, reverse transcription, and cDNA amplification and purification. cDNA production was sheared into 250- to 400-bp short fragments, and an indexed sequencing library was constructed according to the manufacturer’s protocol. Qualification was performed with Qubit ssDNA Assay Kit (Thermo Fisher Scientific Inc.) read using the Agilent Bioanalyzer 2100 (Agilent Technologies). All libraries were sequenced using the DNBSEQ-T7 (MGI) sequencing platform with pair-end sequencing. The sequencing reads contained 30-bp read 1 [including the 10-bp cell barcode 1, 10-bp cell barcode 2, and 10-bp unique molecular identifiers (UMI)] and 100-bp read 2 for gene sequences and 10-bp barcode read for the sample index.

scRNA-seq data processing (alignment, barcode assignment, and UMI counting)

The sequencing data were processed using an open-source pipeline (https://github.com/MGI-techbioinformatics/DNBelab_C_Series_scRNA-analysis-software). Briefly, all samples were performed sample de-multiplexing, barcode processing, and single-cell 3′ UMI counting with default parameters. Processed reads were then aligned to GRCh38 genome reference using STAR (2.5.1b, RRID: SCR_004463; ref. 40). Valid cells were automatically identified based on the UMI number distribution of each cell using the “barcodeRanks()” function of the DropletUtils tool to remove background beads and beads with UMI counts less than the threshold value. Finally, PISA was used to calculate the gene expression of cells and to create a gene × cell matrix for each library.

Unsupervised clustering and cell-type annotation

Cell clustering was conducted using the Seurat (v4.3; ref. 41) package in R software (R version 4.0.4). Genes expressed in less than 3 cells were filtered out, and cells with fewer than 200 genes were excluded. The quality of cells was assessed based on two metrics: (i) the number of detected genes was above 200 and below 5,000; (ii) the percentage of mitochondrial genes was below 25. The libraries were then integrated using the “Merge” and “harmony” functions, and the batch effect was checked if the cells were separately distributed using the “DimPlot” function. The integrated data were then scaled to calculate the principal component (PC) analysis. The first 50 PCs were used to construct the share nearest neighbor network, and the graph-based clustering method Louvain algorithm was used to identify the cell clusters with a resolution of 1.9 across highly variable genes (4,000). Finally, Uniform Manifold Approximation and Projection (UMAP) was used to visualize the clustering results in two-dimensional space. To annotate each cluster as a specific cell type, we selected some classic markers of immune cells, Schwann stromal cells, neuroendocrine cells, neuroblasts, and fibroblasts (4245). The cell types were annotated using a dot plot. The detailed gene list can be found in Supplementary Table S2.

scRNA-seq analysis strategies

The average expression levels of CD112 and CD155 (using “AverageExpression”), as well as the ratios, in tumor cells were calculated, and neuroblastoma tumor samples were divided into high-ratio and low-ratio groups accordingly. DEGs in the given cell types between high-ratio and low-ratio groups or among healthy control (HC) PB, neuroblastoma PB, and neuroblastoma tumor groups were determined using the “FindAllMarkers” function from the Seurat package (one-tailed Wilcoxon rank-sum test). For computing DEGs, all genes were probed provided they were expressed in at least 10% of cells in either of the two populations compared and the expression difference on a natural log scale was at least 0, P value < 0.05. The numbers of DEGs in each comparison were shown as follows: 809 upregulated and 1,840 downregulated genes in γδT cells between HC PB and neuroblastoma PB groups; 2,732 upregulated and 850 downregulated genes in γδT cells between neuroblastoma PB and neuroblastoma tumor groups; 1,850 upregulated and 593 downregulated genes in γδT cells between high-ratio and low-ratio groups; 1,344 upregulated and 942 downregulated genes in CD8 T cells between high-ratio and low-ratio groups; 3,698 upregulated and 3,624 downregulated genes in tumor cells between high-ratio and low-ratio groups; and 1,820 upregulated and 2,622 downregulated genes between high-ratio and low-ratio tumor cells in spatial transcriptomic analysis. Using the upregulated or downregulated DEGs, KEGG pathways enriched in each group were analyzed using “enrichKEGG.” To measure the proliferation status of different cell types, we scored cells using “CellCycleScoring” with a set of characteristic genes involved in the cell cycle (Supplementary Table S2; ref. 46). Gene set variation analysis (GSVA) scores of various cell types were generated using the “GSVA” package with neuroblastoma-related genes or T-cell function–promoting genes or T-cell function–inhibitory genes (Supplementary Table S2). We applied “monocle” to order γδT cells in pseudotime to indicate their developmental trajectories. To visualize the ordered γδT cells in the trajectory, we used the plot_cell_trajectory function to plot the minimum spanning tree on the cells. The starting point of the pseudotime trajectory was determined based on the preliminary understanding of the cell populations used in the analysis. BEAM was used for further analysis of γδT-cell signatures in different fate branches.

BGI STOmics for spatial transcriptome

Tissue processing

Tissue sections were adhered to the Stereo-seq chip (generated by BGI) surface and incubated at 37°C for 3 minutes. Then the sections were fixed in methanol and incubated for 40 minutes at −20°C before Stereo-seq library preparation. Where indicated, the same sections were stained with nucleic acid dye (Thermo Fisher Scientific Inc.), and imaging was performed using a Motic Custom PA53-FS6 microscope prior to in situ capture at FITC channel.

In situ reverse transcription

After washed with 0.1× saline sodium citrate buffer (Thermo Fisher Scientific Inc.) supplemented with 0.05 U/mL RNase inhibitor (NEB), tissue sections placed on the chip were permeabilized using 0.1% pepsin (Sigma-Aldrich) in 0.01 mol/L HCl buffer, incubated at 37°C for 5 minutes before washed with 0.1× SSC buffer supplemented with 0.05 U/mL RNase inhibitor. RNA released from the permeabilized tissue and captured by the DNA nanoball was reverse-transcribed overnight at 42°C using SuperScript II (Invitrogen; 10 U/mL reverse transcriptase, 1 mmol/L dNTPs, 1 mol/L betaine solution PCR reagent, 7.5 mmol/L MgCl2, 5 mmol/L dithiothreitol (DTT), 2 U/mL RNase inhibitor, 2.5 mmol/L Stereo-seq-template switch oligo (TSO), and 1× First-Strand buffer). After reverse transcription, tissue sections were washed twice with 0.1× SSC buffer and digested with Tissue Removal buffer (10 mmol/L Tris-HCl, 25 mmol/L EDTA, 100 mmol/L NaCl, and 0.5% SDS) at 55°C for 10 minutes. cDNA-containing chips were then subjected to Prepare cDNA Release Mix (cDNA Release Enzyme and cDNA Release buffer) treatment at 55°C overnight. cDNA was then purified using the VAHTSTM DNA Clean Beads (0.8×).

Amplification

The resulting cDNAs were amplified with KAPA HiFi HotStart ReadyMix (Roche) with 0.8 mmol/L cDNA-PCR primer. PCRs were conducted as follows: incubation at 95°C for 5 minutes; 15 cycles at 98°C for 20 seconds, 58°C for 20 seconds, and 72°C for 3 minutes; and a final incubation at 72°C for 5 minutes.

Library construction and sequencing

As previously described [Preprint available at Research Square (https://doi.org/10.21203/rs.3.rs-3714208/v1)], the concentrations of the resulting PCR products were quantified using Qubit dsDNA Assay Kit (Thermo Fisher Scientific Inc.). A measure of 20 ng DNA was fragmented with in-house Tn5 transposase at 55°C for 10 minutes, after which, the reaction was stopped by adding 0.02% SDS and gently mixing at 37°C for 5 minutes. Fragmented products were amplified as follows: 25 mL of fragmentation product, 1× KAPA HiFi HotStart Ready Mix, and 0.3 mmol/L Stereo-seq-Library-F primer, and 0.3 mmol/L Stereo-seq-Library-R primer in a total volume of 100 mL with the addition of nuclease-free H2O. The reaction was then run as follows: 1 cycle at 95°C for 5 minutes; 13 cycles at 98°C for 20 seconds, 58°C for 20 seconds, and 72°C for 30 seconds; and 1 cycle at 72°C for 5 minutes. PCR products were purified using the AMPure XP Beads (0.63 and 0.153), used for DNA nanoball generation, and finally sequenced on an MGI DNBSEQ-Tx sequencer.

Stereo-seq raw data processing

FASTQ files were generated using an MGI DNBSEQ-Tx sequencer. Read 1 contained coordinate identity (CID; 1–25 bp) and molecular identity (MID; 26–35 bp), whereas read 2 consisted of the cDNA sequences. CID sequences on the first reads were first mapped to the designed coordinates of the in situ captured chip achieved from the first round of sequencing, allowing one base mismatch to correct for sequencing and PCR errors. Reads with MID containing either N bases or more than two bases with a quality score lower than 10 were filtered out. CID and MID associated with each read were appended to each read header. Retained reads were then aligned to the reference genome (GRCh38_dna_primary_assembly_93) using STAR. Mapped reads with MAPQ >10 were counted and annotated to their corresponding genes. UMI with the same CID and the same gene locus were collapsed, allowing one mismatch to correct for sequencing and PCR errors. Finally, this information was used to generate a CID-containing expression profile matrix. The whole procedure was integrated into a publicly available pipeline SAW available at https://github.com/BGIResearch/SAW.

Spatial transcriptomic analysis strategies

The spatial transcriptomic analysis in this study was conducted according to the operating manual of the STOmics Cloud system (https://en.stomics.tech/). In this system, the technology and products are presented directly in the form of visualization, and the STOmics data are explored and mined in depth through various tools and algorithms. “Bin Size = 50” was used for subsequent analysis. After quality control of data, analytical tool “Spateo” and Louvain algorithm were used to identify the cell clusters with the first 50 PCs and a resolution of 2.0 across highly variable genes (3000). Dimensional reduction was performed using UMAP and “Neighborhood Size = 5.” Classic marker genes, as well as CD112 and CD155, were used to annotate cell clusters as high-ratio tumor cells, low-ratio tumor cells, lymphoid cells, myeloid cells, stromal cells, myocytes, erythrocytes, and mast cells (4245, 47, 48). The detailed gene list can be found in Supplementary Table S2. Using the methods in “scRNA-seq analysis strategies,” DEGs between high-ratio and low-ratio tumor cells were determined, based on which KEGG pathways enriched in each group were analyzed.

Spatial metabolomic analysis

The adjacent tumor tissue sections to that used in spatial transcriptomic analysis were thaw-mounted on a positively charged desorption plate (Thermo Fisher Scientific Inc.). Sections were stored at −80°C before further analysis. They were desiccated at −20°C for 1 hour and then at room temperature for 2 hours before mass spectrometry imaging (MSI) analysis. Meanwhile, an adjacent slice was left for hematoxylin and eosin (H&E) staining. The analyses were performed as previously reported (49). As previously described (50), this experiment was carried out using an AFADESI-MSI platform (Beijing Victor Technology Co., Ltd.) in tandem with a Q-Orbitrap mass spectrometer (Q Exactive, Thermo Fisher Scientific Inc.). Here, the solvent formula was acetonitrile/H2O (8:2) at negative mode and acetonitrile/H2O (8:2) at positive mode, and the solvent flow rate was 1.5 μL/minute, the transporting gas flow rate was 45 L/minute, the spray voltage was set at 7 kV, and the distance between the sample surface and the sprayer was 3 mm as was the distance from the sprayer to the ion-transporting tube. The mass spectrometry (MS) resolution was set at 20,000, the mass range was 70 to 1,200 Da, the automated gain control target was 2E6, the maximum injection time was set to 200 ms, the S-lens voltage was 55 V, and the capillary temperature was 350°C. The MSI experiment was carried out with a constant rate of 0.2 mm/second continuously scanning the surface of the sample section in the x direction and a 100-μm vertical step in the y direction.

The collected .raw files were converted into .imzML format using imzMLConverter (51) and then imported into MSiReader (an open-source interface to view and analyze high resolving power MS imaging files on the MATLAB platform) for ion image reconstructions after background subtraction using the “Cardinal” software package (52). All MS images were normalized using total ion count normalization in each pixel (53). Region-specific MS profiles were precisely extracted by matching high-spatial resolution H&E images. The discriminating endogenous molecules of different tissue microregions were screened using a supervised statistical analytic method: orthogonal partial least squares discrimination analysis (OPLS-DA). Variable Importance in Projection (VIP) values obtained from the OPLS-DA model were used to rank the overall contribution of each variable to group discrimination. The VIP value reflects the importance degree on the classification of sample categories with respect to the first two PCs of the OPLS-DA model, which indicates that this variable has a significant effect if the VIP is greater than 1. A two-tailed Student t test was further used to verify whether the metabolites of difference between groups were significant. Differential metabolites were selected with VIP values greater than 1.0 and P values less than 0.05.

The ions detected by Air Flow Assisted Desorption Electrospray Ionization (AFADESI) were annotated using the PySM pipeline (54) and an in-house SmetDB database (Lumingbio).

Statistical analysis for experimental data

Statistical comparisons were performed using GraphPad Prism software (version 8.0; GraphPad Software Inc., RRID: SCR_002798) or R software (R version 4.0.4). A Student t test was used to analyze the experimental data. Error bars represent the SEM. Significant differences between groups are represented by *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

Data availability

The data analyzed in this study were obtained from GEO at GSE49711 and GSE25624. The raw FASTQ files generated in this study have been deposited in the NCBI Sequence Read Archive database under the accession code SRP537135 (bulk RNA-seq) and SRP539643 (scRNA-seq). All other raw data are available upon request from the corresponding author.

DNAM-1 reduction in functionally impaired γδT cells is related to direct cell-to-cell contact with neuroblastoma tumor cells

To characterize γδT cells in neuroblastomas, we performed scRNA-seq analysis using 21 primary tumor samples, as well as circulating T cells sorted from the PB of HC (n = 3) and patients with neuroblastoma (n = 6; Supplementary Fig. S1A–S1E). After quality control, we analyzed 256,849 cells using UMAP to visualize the cell clusters. Eleven main cell types, including tumor, endothelial, fibroblast, Schwann, myocyte, CD4 T, CD8 T, γδT, NK, B, and myeloid cells, were annotated (Fig. 1A). The fraction of tumor-infiltrated γδT cells markedly increased compared with that of circulating γδT cells (Fig. 1B). Top genes in γδT cells from different groups (HC PB, neuroblastoma PB, and neuroblastoma tumor) and key genes related to T-cell function are illustrated in Fig. 1C. The data indicate that circulating γδT cells exhibit more potent tumor-killing ability than tumor-infiltrating γδT cells, which display exhaustion gene signatures. Functional characteristics of γδT cells were further investigated via KEGG analysis of screened DEGs. KEGG results revealed that circulating γδT cells from patients with neuroblastoma exhibited impaired T-cell activation, poor cytotoxicity, and reduced proliferation compared with those from healthy controls (Fig. 1D). Notably, T-cell activation and cytotoxicity further decreased in tumor-infiltrated γδT cells. Given our previous findings about DNAM-1 reduction in γδT cells from patients with neuroblastoma (13), we analyzed DNAM-1 expression and exhaustion markers (PD-1, TIM3, and TIGIT) across in γδT cells from different groups. Heatmaps and dot plots demonstrated progressively decreased DNAM-1 and increased PD-1 and TIGIT in γδT cells (Fig. 1C and E).

Figure 1.

Accumulated γδT cells in neuroblastoma tumors are featured with reduced DNAM-1 and impaired function revealed by scRNA-seq. A, UMAP plots showing the major cell types and markers. B, Proportions of neuroblastoma tumor-infiltrating immune cells revealed by scRNA-seq. C, Heatmaps showing top 20 marker genes, as well as significant functional genes, in γδT cells from HC PB, NB PB, and NB tumor groups. D and E, Enriched KEGG pathways and expression of TIGIT, HAVCR2, PDCD1, and CD226 in γδT cells from different groups. The dot size in KEGG enrichment analysis represents enriched gene numbers. The results are expressed as the mean ± SEM. Significant differences between groups are represented by *, P < 0.05.

Figure 1.

Accumulated γδT cells in neuroblastoma tumors are featured with reduced DNAM-1 and impaired function revealed by scRNA-seq. A, UMAP plots showing the major cell types and markers. B, Proportions of neuroblastoma tumor-infiltrating immune cells revealed by scRNA-seq. C, Heatmaps showing top 20 marker genes, as well as significant functional genes, in γδT cells from HC PB, NB PB, and NB tumor groups. D and E, Enriched KEGG pathways and expression of TIGIT, HAVCR2, PDCD1, and CD226 in γδT cells from different groups. The dot size in KEGG enrichment analysis represents enriched gene numbers. The results are expressed as the mean ± SEM. Significant differences between groups are represented by *, P < 0.05.

Close modal

We subsequently conducted a thorough immune phenotypic characterization of circulating and tumor-infiltrated γδT cells (Supplementary Fig. S2A and S2B). Along with DNAM-1 reduction (Fig. 2A), γδT cells exhibited increased PD-1, TIM3, and TIGIT levels in PB (Fig. 2B–D), and the majority were effector memory (CD27CD45RA) or central memory (CD27+CD45RA) phenotypes, rather than naïve (CD27+CD45RA+) or terminally differentiated effector memory (CD27CD45RA+) phenotypes (Fig. 2E). Other immune inhibitory molecules, such as CD112R and CD96, were minimally detectable or unaltered (Supplementary Fig. S2C and S2D). Tumor-infiltrated CD45+ immune cells, consisting of T cells, B cells, NK cells, and macrophages, had an increased proportion of γδT cells compared with PB (Fig. 2F), confirming the importance of γδT cells in neuroblastomas. Tumor-infiltrating γδT cells were functionally impaired, with elevated expression of immune inhibitory molecules, including PD-1, TIM3, and TIGIT. The expression of DNAM-1, IFNγ, perforin, and granzyme B (GZMB) was further reduced in tumor-infiltrating γδT cells compared with circulating γδT cells, and their expression levels were already lower than those in HC (Fig. 2G–M; ref. 13). We compared DNAM-1 expression in γδT cells from metastatic BM and blood counterparts from the same patients. γδT cells in metastatic tumors exhibited further DNAM-1 reduction compared with blood counterparts, whereas DNAM-1 expression in BM from patients without metastasis remained unchanged (Fig. 2N–P). Thus, direct contact with tumor cells seems essential for DNAM-1 reduction in γδT cells. Public GSE25624 dataset analysis confirmed DNAM-1 reduction specifically in BM samples from patients with metastasis (Fig. 2Q). To further delineate the necessity of direct cell contact for DNAM-1 regulation, we cocultured in vitro expanded γδT cells (13) with SH-SY5Y cells. Direct coculture induced significant DNAM-1 reduction in γδT cells, whereas a Transwell system preventing direct contact did not (Fig. 2R). Similar results were observed with other three neuroblastoma cell lines (SK-N-BE2, CHLA-255, and IMR-32), albeit to varying extents (Fig. 2S). Additionally, a greater reduction in DNAM-1 was observed with increasing SH-SY5Y cell numbers, suggesting that DNAM-1 reduction correlates with tumor cell quantity (Fig. 2T).

Figure 2.

Circulating and intratumoral γδT cells in patients with neuroblastoma exhibit DNAM-1 reduction and functional exhaustion upon direct cell contact with neuroblastoma cells. A–D, Expression of DNAM-1, PD-1, TIM3, and TIGIT in PB γδT cells from HC and neuroblastoma (NB). E, Cell compositions of circulating γδT cells [naïve (Tnaïve), central memory (TCM), effector memory (TEM), terminally differentiated effector memory (TEMRA)]. F, Immune fractions in primary neuroblastoma tumors. G–M, Expression of DNAM-1, PD-1, TIM3, TIGIT, IFNγ, perforin, and GZMB in neuroblastoma circulating γδT cells and tumor-infiltrating γδT cells. N–P, DNAM-1 expression in neuroblastoma γδT cells from PB and corresponding BM. MFI, mean fluorescence intensity. Q, DNAM-1 expression analysis using the GSE25624 dataset of neuroblastoma BM. R, DNAM-1 expression analysis in the coculture system of SH-SY5Y cells and in vitro expanded γδT cells with (left) or without (right) a Transwell chamber. S, Effects of neuroblastoma cell lines on DNAM-1 reduction in γδT cells during coculture. T, Effects of SH-SY5Y at different ratios on DNAM-1 reduction in γδT cells. U, Cytotoxicity assays using HC or neuroblastoma γδT cells in the presence of α-DNAM-1. V, Using the GSE49711 dataset, the expression pattern of DNAM-1 was analyzed in neuroblastoma tumors (n = 498) of different groups. amp, amplification; HR, high risk; LR, low risk; MR, intermediate risk; noamp, no amplification. W, Overall and event-free survival curves were generated by grouping samples with the median of DNAM-1 expression. The results are expressed as the mean ± SEM from at least three independent experiments. Significant differences between groups are represented by ns, not significant; *, P < 0.05; **, P < 0.01; ***; P < 0.001; ****, P < 0.0001.

Figure 2.

Circulating and intratumoral γδT cells in patients with neuroblastoma exhibit DNAM-1 reduction and functional exhaustion upon direct cell contact with neuroblastoma cells. A–D, Expression of DNAM-1, PD-1, TIM3, and TIGIT in PB γδT cells from HC and neuroblastoma (NB). E, Cell compositions of circulating γδT cells [naïve (Tnaïve), central memory (TCM), effector memory (TEM), terminally differentiated effector memory (TEMRA)]. F, Immune fractions in primary neuroblastoma tumors. G–M, Expression of DNAM-1, PD-1, TIM3, TIGIT, IFNγ, perforin, and GZMB in neuroblastoma circulating γδT cells and tumor-infiltrating γδT cells. N–P, DNAM-1 expression in neuroblastoma γδT cells from PB and corresponding BM. MFI, mean fluorescence intensity. Q, DNAM-1 expression analysis using the GSE25624 dataset of neuroblastoma BM. R, DNAM-1 expression analysis in the coculture system of SH-SY5Y cells and in vitro expanded γδT cells with (left) or without (right) a Transwell chamber. S, Effects of neuroblastoma cell lines on DNAM-1 reduction in γδT cells during coculture. T, Effects of SH-SY5Y at different ratios on DNAM-1 reduction in γδT cells. U, Cytotoxicity assays using HC or neuroblastoma γδT cells in the presence of α-DNAM-1. V, Using the GSE49711 dataset, the expression pattern of DNAM-1 was analyzed in neuroblastoma tumors (n = 498) of different groups. amp, amplification; HR, high risk; LR, low risk; MR, intermediate risk; noamp, no amplification. W, Overall and event-free survival curves were generated by grouping samples with the median of DNAM-1 expression. The results are expressed as the mean ± SEM from at least three independent experiments. Significant differences between groups are represented by ns, not significant; *, P < 0.05; **, P < 0.01; ***; P < 0.001; ****, P < 0.0001.

Close modal

We have reported that the immunocompetence of γδT cells depends on DNAM-1 (13). Our coculture experiments demonstrated that γδT cytotoxicity could be abolished by adding DNAM-1–neutralizing antibodies, and γδT cells from patients with neuroblastoma exhibited reduced tumoricidal ability (Fig. 2U). Bioinformatic analysis using the GSE49711 dataset revealed a strong correlation between DNAM-1 expression and neuroblastoma prognosis: stage IV tumors, high-risk cases, MYCN amplification, and patient death were all associated with reduced DNAM-1 expression, and low DNAM-1 expression correlated with poor survival probability (Fig. 2V and W).

These data collectively demonstrate that neuroblastoma tumor cells downregulate DNAM-1 expression and impair γδT-cell function through direct cell-to-cell contact.

Tumor cells differentially expressing DNAM-1 ligands CD112 and CD155 belong to distinct functional subsets and display locale-specific patterns

Expanding on our findings about DNAM-1, we examined the expression of its ligands, CD112 and CD155, in tumor cells. Expression levels of these ligands in various tumor cells, including primary neuroblastoma cells, were significantly different (Supplementary Fig. S3A–S3J). We examined whether there was a disparity in the spatial distribution of CD112 and CD155 within the tumor. Spatial transcriptomic analysis was performed using tissue sections from a high-risk neuroblastoma tumor to generate a transcriptomic landscape of CD112 and CD155 expression in the tumor tissue. Following H&E staining, bright-field imaging, sequencing, mapping, filtering, and annotation (Supplementary Fig. S4A–S4C), 71,364 cells were validated, and these were clustered into eight main cell types: tumor cells expressing a high CD112/CD155 ratio, low-ratio tumor cells, lymphoid cells, myeloid cells, stromal cells, myocytes, erythrocytes, and mast cells (Fig. 3A). The spatial transcriptomic data revealed a locale-specific pattern in which CD155 was enriched in a particular sublocale, whereas CD112 transcription was evenly distributed across the entire tumor mass (Fig. 3B). This spatial expression pattern was confirmed by IHC staining with antibodies against CD155 and CD112. Tissue sections of primary tumors displayed a partitioned pattern of CD155 expression alongside evenly distributed CD112 expression, which was absent in the paratumor regions (Supplementary Fig. S5). KEGG pathway analysis showed that CD112/CD155 high-ratio tumor cells were enriched with neuronal functions, suggesting a more differentiated state. In contrast, low-ratio tumor cells were enriched in pathways related to self-renewal and metabolism, implying increased proliferation and metabolic changes (Fig. 3C). Consistently, genes implicated in cell proliferation, including RRM2, MKI67, TOP2A, and CENPF, as well as the cell-cycle score, were higher in low-ratio tumor cells compared with high-ratio tumor cells (Supplementary Fig. S6A and S6B).

Figure 3.

Tumors with different CD112/CD155 expression ratios belong to distinct functional subsets with featured gene expression profiles and different functional γδT-cell infiltrations. A, Major cell types arranged in the space of the sample tissue section and H&E image. B, Spatial distribution and expression of CD112 and CD155. C, Enriched KEGG pathways in high-ratio and low-ratio tumor cells. D, Violin plots showing the expression levels of CD112 and CD155, as well as CD112/CD155 ratios, in tumor cells. E, Violin plots showing the proportions of various immune cells among all immune cells in high-ratio and low-ratio groups. F and G, Enriched KEGG pathways in tumor cells and γδT cells from high-ratio and low-ratio groups. The dot size in KEGG enrichment analysis represents enriched gene numbers. The results are expressed as the mean ± SEM. Significant differences between groups are represented by ns, not significant; *, P < 0.05.

Figure 3.

Tumors with different CD112/CD155 expression ratios belong to distinct functional subsets with featured gene expression profiles and different functional γδT-cell infiltrations. A, Major cell types arranged in the space of the sample tissue section and H&E image. B, Spatial distribution and expression of CD112 and CD155. C, Enriched KEGG pathways in high-ratio and low-ratio tumor cells. D, Violin plots showing the expression levels of CD112 and CD155, as well as CD112/CD155 ratios, in tumor cells. E, Violin plots showing the proportions of various immune cells among all immune cells in high-ratio and low-ratio groups. F and G, Enriched KEGG pathways in tumor cells and γδT cells from high-ratio and low-ratio groups. The dot size in KEGG enrichment analysis represents enriched gene numbers. The results are expressed as the mean ± SEM. Significant differences between groups are represented by ns, not significant; *, P < 0.05.

Close modal

To verify the difference between tumor cells with low and high CD112/CD155 expression ratios, scRNA-seq data from primary neuroblastoma tumors in 21 patients were analyzed (Fig. 3D). Neuroblastoma tumor samples were divided into high-ratio (NBT1, 4, 5, 7, 9, 10, 11, 12, 14, 17, and 21) and low-ratio (NBT2, 3, 6, 8, 13, 15, 16, 18, 19, and 20) groups using the median expression ratios of CD112 versus CD155. Single-cell sequencing analysis revealed that high-ratio neuroblastoma tumors were associated with fewer total T cells but higher numbers of myeloid and γδT cells among infiltrated immune cells (Fig. 3E). Subsequently, enriched pathways and the functional characteristics of tumor, γδT, CD8 T, and NK cells were examined using DEGs between the two ratio groups. Similar to the spatial transcriptomic data, scRNA-seq showed enriched neuronal function pathways in high-ratio tumor cells and enriched self-renewal and different metabolic pathways in low-ratio tumor cells (Fig. 3F). Cell-cycle score analysis further indicated that low-ratio tumor cells had stronger proliferative activity and a gene expression profile associated with neuroblastoma (Supplementary Fig. S6C and S6D).

We performed COMPASS analysis (55) using scRNA-seq data and spatial metabolomic analysis with adjacent tumor tissue sections to characterize the metabolic states of tumor cells. Supplementary Figure S7A and S7B indicate enhanced tyrosine and phenylalanine metabolism in high-ratio tumor cells, compared with robust lipid metabolism in low-ratio tumor cells. Changes in metabolism were confirmed using representative metabolites in different groups, such as enriched fatty acids in low-ratio tumor cells and enriched pyruvic acid in high-ratio tumor cells (Supplementary Fig. S7C).

With regard to cytotoxic immune cells, scRNA-seq data showed an enhanced gene expression profile related to T-cell response, self-renewal, and activation signaling pathways in high-ratio tumors (Fig. 3G; Supplementary Figs. S6E–S6G and S8A–S8H). Furthermore, γδT cells in high-ratio tumors exhibited higher proliferation and increased IFNG and PRF1 expression.

Collectively, our results reveal the intertumoral and intratumoral heterogeneity of CD112 and CD155 expression. Tumor cells with different CD112/CD155 expression ratios belong to distinct functional subsets with varying differentiation statuses, metabolic profiles, and proliferation rates, and they are associated with differential immune cell infiltration and activation.

Differential expression of CD112 and CD155 on tumor cells modulates γδT cell–mediated killing and correlates with patient prognosis

Considering the heterogeneity in the expression levels and spatial distribution of CD112 and CD155 in neuroblastoma tumors, we hypothesized that different ratios of these two molecules could represent the risk of tumor cells and modulate γδT-cell responses. Analysis of the neuroblastoma GSE49711 dataset revealed that neuroblastoma tumors with high risk, MYCN amplification, and death outcomes presented with lower CD112/CD155 expression ratios (Fig. 4A). We then segregated neuroblastoma samples into two subgroups with high and low CD112/CD155 ratios and found that patients with neuroblastoma with high CD112/CD155 ratios had better survival rates (Fig. 4B). Based on the gene expression patterns between these two groups, the expression of cytolytic factors, such as IFNγ, perforin, and GZMB, was enhanced in the high-ratio group, similar to DNAM-1 (Fig. 4C).

Figure 4.

CD112/CD155 expression ratios in NB are associated with prognosis of patients with neuroblastoma (NB) and direct γδT-cell functional fate. A, Using the GSE49711 dataset, CD112/CD155 expression ratios were analyzed in NB tumors of different groups. amp, amplification; noamp, no amplification. B, Overall and event-free survival curves were generated by grouping samples with the median of CD112/CD155 expression ratios. C, T-cell antitumor function–related genes were analyzed between high- and low-ratio groups. D–F, The CD112/CD155 ratios in different tumor groups were calculated according to the mean fluorescence intensity (MFI) detected by FACS. G–I, DNAM-1, PD-1, and TIM3 expression was detected in γδT cells cocultured with SK-N-BE2 cells pretreated with doxycycline. J and K, γδT-cell cytotoxicity against SK-N-BE2 cells pretreated with doxycycline. The results are expressed as the mean ± SEM from at least three independent experiments. Significant differences between groups are represented by ns, no significance; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. HR, high risk; LR, low risk; MR, intermediate risk.

Figure 4.

CD112/CD155 expression ratios in NB are associated with prognosis of patients with neuroblastoma (NB) and direct γδT-cell functional fate. A, Using the GSE49711 dataset, CD112/CD155 expression ratios were analyzed in NB tumors of different groups. amp, amplification; noamp, no amplification. B, Overall and event-free survival curves were generated by grouping samples with the median of CD112/CD155 expression ratios. C, T-cell antitumor function–related genes were analyzed between high- and low-ratio groups. D–F, The CD112/CD155 ratios in different tumor groups were calculated according to the mean fluorescence intensity (MFI) detected by FACS. G–I, DNAM-1, PD-1, and TIM3 expression was detected in γδT cells cocultured with SK-N-BE2 cells pretreated with doxycycline. J and K, γδT-cell cytotoxicity against SK-N-BE2 cells pretreated with doxycycline. The results are expressed as the mean ± SEM from at least three independent experiments. Significant differences between groups are represented by ns, no significance; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. HR, high risk; LR, low risk; MR, intermediate risk.

Close modal

We further collected 38 neuroblastoma tissue samples for bulk RNA-seq. Consistently, immune activation signals were found to be associated with high-ratio tumors, whereas low-ratio tumors were associated with genes related to robust neuroblastoma tumor growth and distinct metabolic activities (Supplementary Fig. S9A–S9H).

Next, we measured the CD112/CD155 ratios in primary neuroblastoma tumor cells to verify the bioinformatic analysis of the public database. FACS data indicated that the ratio of CD112/CD155 expression (mean fluorescence intensity) was markedly lower in tumors of stage IV patients relative to lower INSS stages, in high-risk tumors compared with low-risk tumors and in liver metastatic neuroblastoma tumors compared with orthotopic neuroblastoma tumors (Fig. 4D–F). To mimic tumors expressing different ratios of CD112/CD155, we constructed a Tet-on controllable expression system in SK-N-BE2 cells, in which the upregulation of CD155 and downregulation of CD112 were simultaneously attained (SK-N-BE2-CD155-shCD112 cells; Supplementary Fig. S10A and S10B). Using this design, we created tumor cells with higher CD112/CD155 expression ratios in the presence of lower concentrations of doxycycline and those with lower ratios with higher concentrations of doxycycline. We cocultured the SK-N-BE2-CD155-shCD112 cells with γδT cells in the presence of different concentrations of doxycycline. Notably, suppression of DNAM-1 expression in γδT cells gradually increased as the doxycycline concentration increased, suggesting that tumor cells with a lower CD112/CD155 ratio induced a greater reduction in DNAM-1 expression (Fig. 4G). Increased expression of the T-cell exhaustion markers PD-1 and TIM3 occurred along with DNAM-1 reduction after coculture with tumor cells expressing a lower CD112/CD155 ratio (Fig. 4H and I). Accordingly, tumor cells with a lower ratio exhibited higher resistance to γδT cell–mediated cytotoxicity in a doxycycline dosage–dependent manner (Fig. 4J). We monitored tumor cell proliferation and γδT cell–mediated tumor cell killing using RTCA. Our results indicated that without changing the tumor proliferation rate, tumor cells with a lower CD112/CD155 expression ratio were more resistant to γδT-cell killing (Fig. 4K).

Certain chemotherapeutic drugs or small-molecule antitumor drugs contribute to T cell– or NK cell–mediated tumor killing by regulating ligand expression of activating receptors on tumor cells (56). Notably, doxorubicin, commonly used in high-risk neuroblastoma treatment (57), induced CD112 upregulation and CD155 downregulation simultaneously, rendering tumor cells more susceptible to γδT-cell killing (Supplementary Fig. S11A–S11H).

Bioinformatic data from a public database, direct measurement of primary tumors, a cell model mimicking tumors with differential ratios of CD112/CD155 expression, and the effect of chemotherapy drug on these two ligands strongly suggested that changes in the CD112/CD155 ratio could effectively modulate tumor resistance or susceptibility to γδT cells, ultimately leading to different disease outcomes.

The presence of CD112 on tumor cells is essential for sustaining DNAM-1–mediated γδT-cell activation

We established that tumor cell heterogeneity in CD112 versus CD155 expression ratios represented distinct functional subsets, predicting different disease outcomes through modulating γδT-cell tumoricidal activity and DNAM-1 expression levels. To investigate the correlation of CD112 and CD155 with DNAM-1 expression and γδT-cell activation, we analyzed whether CD112 exhibited any expression pattern corresponding to changes in tumor risk. In the GSE49711 dataset, high-risk neuroblastoma tumors showed significantly lower levels of CD112 compared with those with low or intermediate risk. However, CD112 expression did not vary significantly with different INSS stages, MYCN amplification statuses, or patient mortality (Fig. 5A). Additionally, survival analysis indicated that neuroblastoma prognosis was poorly associated with CD112 levels in patients (Fig. 5B). Bioinformatic analysis revealed a positive correlation between CD112 and DNAM-1 levels in patients with neuroblastoma (Fig. 5C). CD112 did not regulate DNAM-1 expression (similar results were observed for TIGIT, CD96, and CD112R; Supplementary Fig. S12A–S12D) as rhCD112 treatment or CD112-neutralizing antibody (α-CD112) in coculture with neuroblastoma tumor cells did not significantly impact DNAM-1 levels in γδT cells (Fig. 5D–F). Nevertheless, the direct addition of rhCD112 in coculture promoted γδT cell–mediated tumor killing, whereas the addition of α-CD112 inhibited it (Supplementary Fig. S12E–S12L). To further validate these findings, we constructed CHLA-255 neuroblastoma tumor cells with stable CD112 knockdown (CHLA-255-shCD112), which exhibited enhanced cell migration (Fig. 5G; Supplementary Fig. S12M–S12O), and IMR-32 neuroblastoma tumor cells (which express relatively low levels of CD112) overexpressing CD112 (IMR-32-CD112; Supplementary Fig. S12P). Neither CD112 knockdown nor overexpression affected DNAM-1 expression in coculture (Fig. 5H; Supplementary Fig. S12Q). The absence of CD112 rendered tumor cells less susceptible to γδT cell–mediated killing because of weaker activation of γδT cells, whereas CD112 overexpression increased susceptibility to γδT-cell killing (Fig. 5I; Supplementary Fig. S12R and S12S).

Figure 5.

Rather than CD112, CD155 reduces DNAM-1 in γδT cells and suppresses γδT-cell cytotoxicity against neuroblastoma. A, Using the GSE49711 dataset, the expression patterns of CD112 were analyzed in neuroblastoma tumors of different groups. B, Overall and event-free survival curves were generated by grouping samples with the median of CD112 expression. C, The expression correlation between CD112 and DNAM-1 was analyzed using the GSE49711 dataset. D and E, DNAM-1 expression was detected upon rhCD112 treatment. F, DNAM-1 reduction was detected in the coculture of γδT cells and SH-SY5Y cells with α-CD112 treatment. G, CD112 was knocked down in CHLA-255 cells stably. H, DNAM-1 reduction was detected after coculture with CHLA-255-NC and CHLA-255-shCD112 cells. I, γδT-cell cytotoxicity against CHLA-255-NC and CHLA-255-shCD112 cells. J, Using the GSE49711 dataset, the expression patterns of CD155 were analyzed in neuroblastoma tumors of different groups. K, Overall and event-free survival curves were generated by grouping samples with the median of CD155 expression. L, The expression correlation between CD155 and DNAM-1 was analyzed using the GSE49711 dataset. M and N, DNAM-1 expression was detected upon rhCD155 treatment. O, DNAM-1 reduction was detected in the coculture of γδT cells and SH-SY5Y cells with α-CD155 treatment. P, CD155 was knocked down in SK-N-BE2 cells stably. Q, DNAM-1 was detected after coculture with SK-N-BE2-NC and SK-N-BE2-KoCD155 cells. R, γδT-cell cytotoxicity against SK-N-BE2-NC and SK-N-BE2-KoCD155 cells. S, The neuroblastoma experimental model was constructed by inoculating SK-N-BE2-KoCD155 cells and SK-N-BE2-NC cells subcutaneously into nude mice. Tumor growth curves were generated by measuring tumor sizes, and tumors were weighed after surgery resection. The infiltration of γδT cells in tumors was detected by IHC. T, Using SK-N-BE2-luc-NC/KoCD155 cells, the orthotropic neuroblastoma tumor model was also constructed. Tumor sizes were monitored using the IVIS Spectrum Imaging System, and the radiance was calculated (blue lines for spliced together images). The results are expressed as the mean ± SEM from at least three independent experiments. Significant differences between groups are represented by ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. amp, amplification; HR, high risk; LR, low risk; MFI, mean fluorescence intensity; MR, intermediate risk; NC, negative control; noamp, no amplification; RQ, relative quantity.

Figure 5.

Rather than CD112, CD155 reduces DNAM-1 in γδT cells and suppresses γδT-cell cytotoxicity against neuroblastoma. A, Using the GSE49711 dataset, the expression patterns of CD112 were analyzed in neuroblastoma tumors of different groups. B, Overall and event-free survival curves were generated by grouping samples with the median of CD112 expression. C, The expression correlation between CD112 and DNAM-1 was analyzed using the GSE49711 dataset. D and E, DNAM-1 expression was detected upon rhCD112 treatment. F, DNAM-1 reduction was detected in the coculture of γδT cells and SH-SY5Y cells with α-CD112 treatment. G, CD112 was knocked down in CHLA-255 cells stably. H, DNAM-1 reduction was detected after coculture with CHLA-255-NC and CHLA-255-shCD112 cells. I, γδT-cell cytotoxicity against CHLA-255-NC and CHLA-255-shCD112 cells. J, Using the GSE49711 dataset, the expression patterns of CD155 were analyzed in neuroblastoma tumors of different groups. K, Overall and event-free survival curves were generated by grouping samples with the median of CD155 expression. L, The expression correlation between CD155 and DNAM-1 was analyzed using the GSE49711 dataset. M and N, DNAM-1 expression was detected upon rhCD155 treatment. O, DNAM-1 reduction was detected in the coculture of γδT cells and SH-SY5Y cells with α-CD155 treatment. P, CD155 was knocked down in SK-N-BE2 cells stably. Q, DNAM-1 was detected after coculture with SK-N-BE2-NC and SK-N-BE2-KoCD155 cells. R, γδT-cell cytotoxicity against SK-N-BE2-NC and SK-N-BE2-KoCD155 cells. S, The neuroblastoma experimental model was constructed by inoculating SK-N-BE2-KoCD155 cells and SK-N-BE2-NC cells subcutaneously into nude mice. Tumor growth curves were generated by measuring tumor sizes, and tumors were weighed after surgery resection. The infiltration of γδT cells in tumors was detected by IHC. T, Using SK-N-BE2-luc-NC/KoCD155 cells, the orthotropic neuroblastoma tumor model was also constructed. Tumor sizes were monitored using the IVIS Spectrum Imaging System, and the radiance was calculated (blue lines for spliced together images). The results are expressed as the mean ± SEM from at least three independent experiments. Significant differences between groups are represented by ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. amp, amplification; HR, high risk; LR, low risk; MFI, mean fluorescence intensity; MR, intermediate risk; NC, negative control; noamp, no amplification; RQ, relative quantity.

Close modal

Collectively, our findings revealed that although CD112 on tumor cells does not regulate DNAM-1 expression, it is essential for γδT-cell activation.

Interacting with CD155 on tumor cells reduces DNAM-1 expression and suppresses γδT-cell cytotoxicity

We hypothesized that CD155 on tumor cells contributed to DNAM-1 reduction in γδT cells upon tumor contact. To test this, we analyzed the GSE49711 dataset and observed a positive correlation between CD155 expression and high-risk tumors, MYCN amplification, and death outcomes (Fig. 5J). Patients with CD155 expression above the median level exhibited significantly lower survival rates (Fig. 5K). Importantly, CD155 expression was inversely correlated with DNAM-1 mRNA levels in patients with neuroblastoma (Fig. 5L). In vitro experiments demonstrated that rhCD155 reduced DNAM-1 expression without affecting other competing receptors, such as CD96, CD112R, and TIGIT (Fig. 5M and N; Supplementary Fig. S13A–S13F). This reduction was accompanied by a marked inhibition of γδT-cell cytotoxicity (Supplementary Fig. S13G–S13Q). Consistently, coculturing SH-SY5Y cells with γδT cells led to DNAM-1 reduction and decreased tumoricidal activity, both of which were reversed by a CD155-neutralizing antibody (α-CD155; Fig. 5O; Supplementary Fig. S14A–S14K).

Next, we established CD155-knockout SK-N-BE2 cells using the CRISPR-Cas9 system (SK-N-BE2-KoCD155; Fig. 5P). SK-N-BE2-KoCD155 cells displayed reduced migration (Supplementary Fig. S13R–S13T) and failed to induce DNAM-1 reduction, promoting stronger γδT-cell activation in coculture (Fig. 5Q; Supplementary Fig. S13U), with only minor effects on TIGIT downregulation (Supplementary Fig. S13V). In the absence of CD155, SK-N-BE2-KoCD155 cells became more susceptible to γδT cell–mediated killing (Fig. 5R).

To validate CD155 function in vivo, we utilized a subcutaneous tumor xenograft model. The effectiveness of γδT cell–mediated antitumor immunity was evidenced by reduced tumor size and weight compared with sham controls (PBS injection). Notably, γδT cells induced greater tumor shrinkage in mice inoculated with SK-N-BE2-KoCD155 cells compared with those inoculated with SK-N-BE2-NC cells. IHC analysis of excised tumors ruled out differential infiltration of γδT cells (Fig. 5S). To mimic immunotherapy in patients with neuroblastoma, we developed an orthotopic neuroblastoma model by implanting SK-N-BE2-luc-NC/KoCD155 cells, which ectopically express luciferase, into the left kidneys of NOD/SCID gamma mice. As expected, γδT cells exhibited enhanced cytotoxic activity against tumors formed from SK-N-BE2-luc-KoCD155 cells compared with SK-N-BE2-luc-NC control tumors (Fig. 5T). Altogether, our findings demonstrate that interaction with CD155 expression on tumor cells induces DNAM-1 reduction and inhibits γδT-cell cytotoxicity.

The TRIM21-mediated ubiquitin proteasome pathway is responsible for γδT-cell DNAM-1 reduction after interacting with CD155 on tumor cells

When we investigated the DNAM-1 mRNA levels in γδT cells treated with rhCD155 (Fig. 6A), we did not perceive a change whatsoever. We hypothesized that DNAM-1 reduction by interacting with CD155 occurs at the translational, but not the transcriptional, level. After direct contact with SK-N-BE2 tumor cells in coculture, both membrane expression and total DNAM-1 expression in γδT cells decreased, consistent with the results from rhCD155 treatment (Fig. 6B). Furthermore, using a confocal microscope, a reduction in DNAM-1 expression on the γδT-cell membrane became noticeable when in proximity to SK-N-BE2-NC cells but not to SK-N-BE2-KoCD155 cells (Fig. 6C).

Figure 6.

CD155 induces DNAM-1 degradation via TRIM21-mediated ubiquitination in γδT cells during the interaction with neuroblastoma cells. A, DNAM-1 expression at the mRNA level was detected in γδT cells treated with rhCD112 or rhCD155. B, The membrane DNAM-1 expression and total DNAM-1 expression at the protein level were detected in γδT cells cocultured with SK-N-BE2 cells. C, The coculture of SK-N-BE2-NC or SK-N-BE2-KoCD155 cells (green) with γδT cells was observed by confocal microscopy to detect DNAM-1 (red) reduction. D–G, DNAM-1 expression was detected in γδT cells cocultured with SK-N-BE2 cells or treated with rhCD155 in the presence of MG132 or NH4Cl. H–K, DNAM-1 expression was detected in γδT cells cocultured with SK-N-BE2 cells or treated with rhCD155 in the presence of PP2. L, Coimmunoprecipitation assay was performed on rhCD155-treated γδT cells and anti–DNAM-1 antibody. M, γδT cells were treated with rhCD155 prior to immunostaining with antibodies against TRIM21 (green) and DNAM-1 (red). N, TRIM21 was knocked down in γδT cells. O and P, DNAM-1 reduction in TRIM21-knockdown γδT cells was detected upon rhCD155 treatment. The results are expressed as the mean ± SEM from at least three independent experiments. Significant differences between groups are represented by ns, not significant; *, P < 0.05; **, P < 0.01. MFI, mean fluorescence intensity; NC, negative control; RQ, relative quantity.

Figure 6.

CD155 induces DNAM-1 degradation via TRIM21-mediated ubiquitination in γδT cells during the interaction with neuroblastoma cells. A, DNAM-1 expression at the mRNA level was detected in γδT cells treated with rhCD112 or rhCD155. B, The membrane DNAM-1 expression and total DNAM-1 expression at the protein level were detected in γδT cells cocultured with SK-N-BE2 cells. C, The coculture of SK-N-BE2-NC or SK-N-BE2-KoCD155 cells (green) with γδT cells was observed by confocal microscopy to detect DNAM-1 (red) reduction. D–G, DNAM-1 expression was detected in γδT cells cocultured with SK-N-BE2 cells or treated with rhCD155 in the presence of MG132 or NH4Cl. H–K, DNAM-1 expression was detected in γδT cells cocultured with SK-N-BE2 cells or treated with rhCD155 in the presence of PP2. L, Coimmunoprecipitation assay was performed on rhCD155-treated γδT cells and anti–DNAM-1 antibody. M, γδT cells were treated with rhCD155 prior to immunostaining with antibodies against TRIM21 (green) and DNAM-1 (red). N, TRIM21 was knocked down in γδT cells. O and P, DNAM-1 reduction in TRIM21-knockdown γδT cells was detected upon rhCD155 treatment. The results are expressed as the mean ± SEM from at least three independent experiments. Significant differences between groups are represented by ns, not significant; *, P < 0.05; **, P < 0.01. MFI, mean fluorescence intensity; NC, negative control; RQ, relative quantity.

Close modal

To understand the molecular mechanism of DNAM-1 degradation, the proteasome inhibitor MG132 or the lysosome inhibitor NH4Cl was added to the medium during γδT-cell expansion, in the presence of rhCD155 treatment, or SK-N-BE2 tumor cell coculture. Both FACS and Western blotting revealed that DNAM-1 degradation could be partially inhibited by MG132, but not by NH4Cl, suggesting that the ubiquitin–proteasome pathway is responsible for CD155-induced DNAM-1 degradation (Fig. 6D–G). Src kinases are indispensable for DNAM-1 degradation by ubiquitination in CD8+ T and NK cells (58, 59). The specific Src kinase inhibitor PP2 prevented γδT cells from DNAM-1 degradation in coculture or rhCD155 treatment (Fig. 6H–K), implying that Src kinase–mediated ubiquitination is involved in CD155-induced DNAM-1 degradation in γδT cells.

A potential E3 ubiquitin ligase is implicated in the ubiquitin–proteasome pathway–dependent CD155-mediated DNAM-1 degradation (60). Subsequently, we performed coimmunoprecipitation with DNAM-1 and ultrahigh-performance LC/MS-MS. The TRIM21 was identified, which mediates protein ubiquitination and degradation during various biological processes, including the antitumor function of cytotoxic T cells (61), antiviral responses, and autoimmune diseases (62). Following rhCD155 treatment, the interaction between DNAM-1 and TRIM21 was verified using immunoprecipitation and Western blotting (Fig. 6L). In addition, the colocalization of DNAM-1 and TRIM21 was observed using confocal microscopy (Fig. 6M). As expected, γδT cells with TRIM21 knocked down (Fig. 6N) were resistant to CD155-induced DNAM-1 degradation (Fig. 6O and P).

Intratumoral γδT cells exhibit varying degrees of DNAM-1 expression and functional activation from interplaying with tumor heterogeneity

We investigated whether γδT cells within tumors comprise different functional subsets as a result of interacting with tumor cells that differentially express DNAM-1 ligands. Data from scRNA-seq revealed that γδT cells consist of three major functional subsets: IFNγ producing (type 1), IL17 producing (type 3), and IFNγ IL17 CXCR6hi DNAM-1lo γδT cells (novel). The novel γδT-cell subpopulation completely lost the expression of IFNG and RORC. They were characterized by very low levels of DNAM-1 and GZMB expression but a considerably high expression of GZMA, GZMK and GZMH, which implies dysregulation in immune activation and cytotoxicity (Fig. 7A–C). Importantly, these three functional γδT subsets were differentially distributed between tumors with high CD112/CD155 ratios and those with low ratios. Although type 1 γδT cells predominated in high-ratio tumors, type 3 γδT cells enriched and novel IFNγ IL17 CXCR6hi DNAM-1lo γδT cells prevailed in low-ratio tumors (Fig. 7A). Pseudotime analysis based on the scRNA-seq data (Fig. 7D) revealed that γδT cells diverged into three distinct differentiation fates, from naïve to fate 1 and fate 2. Type 3 and novel γδT subsets were primarily terminally differentiated cells at fate 1, which were enriched in CD112/CD155 low-ratio tumors. The majority of type 1 γδT cells were fate 2 cells and dominated in high-ratio tumors (Fig. 7D). These results emphasize that heterogeneity in tumor CD112 and CD155 expression configures a differential TME where varying degrees of DNAM-1 expression and γδT-cell activation are achieved through interactions between the immune-activating receptor DNAM-1 and its tumor ligands.

Figure 7.

Neuroblastoma tumor-infiltrating γδT cells display distinct subset characteristics with reduced DNAM-1 and poor cytotoxicity in low-ratio group tumors. A, UMAP plots showing the major γδT-cell types and bar plot showing the proportions of different types of γδT cells in high-ratio and low-ratio groups. B and C, UMAP plots and dot plots showing the expression of signature genes. D, Monocle-guided cell trajectories of γδT cells according to pseudotime and types. The results are expressed as the mean ± SEM. Significant differences between groups are represented by ns, not significant; *, P < 0.05.

Figure 7.

Neuroblastoma tumor-infiltrating γδT cells display distinct subset characteristics with reduced DNAM-1 and poor cytotoxicity in low-ratio group tumors. A, UMAP plots showing the major γδT-cell types and bar plot showing the proportions of different types of γδT cells in high-ratio and low-ratio groups. B and C, UMAP plots and dot plots showing the expression of signature genes. D, Monocle-guided cell trajectories of γδT cells according to pseudotime and types. The results are expressed as the mean ± SEM. Significant differences between groups are represented by ns, not significant; *, P < 0.05.

Close modal

Pediatric solid tumors in general and neuroblastoma in specific are commonly characterized by poor immunogenicity and reduced MHC molecule expression (63, 64). The immunocompetence of γδT cells is usually independent of tumor-associated antigens and MHC expression; these cells are characterized by strong cytotoxicity toward various tumors in experimental and clinical settings (65, 66). Additionally, γδT cells are handily expanded in large quantities in vitro, making them appropriate candidates for immunotherapy (67). Among 22 immune populations across 25 types of human cancers, the intratumoral γδT cell is one of a kind that best associated with a favorable prognosis of cancers (68). This finding is consistent with our data showing that tumor-infiltrating γδT cells are disproportionately enriched in neuroblastoma tumors. Notably, our previous studies have demonstrated that DNAM-1 is indispensable for γδT cell–mediated tumor cytotoxicity (13). Understanding the complex cross-talk between DNAM-1 on γδT cells and their tumor ligands in the TME is, therefore, an important step in improving γδT-cell immunotherapy for tumors.

In the TME, the cellular components comprise tumor cells with heterogeneous fitness, as well as pro- and antitumor immune cells. Their intertwined interplay shapes the function of immune cells and determines the fate of tumor cells. For example, IL22 derived from CD4 T helper cells can promote CD155 expression in tumor cells and consequently suppress DNAM-1 expression in NK cells, leading to tumor metastasis (69). Several checkpoint molecules can be modulated by interactions with their ligands in tumor cells. It is difficult to decipher the complex interactions between CD112 or CD155 and their receptors, including DNAM-1, TIGIT, CD96, and CD112R, owing to the complexity of the molecular cross-talk and signaling feedback. In the present study, we found that CD155 directly downregulates DNAM-1 expression upon ligand–receptor ligation. The DNAM-1–CD155 axis could be a trailblazer in determining tumor fate because the association of CD155 with patient prognosis is staggeringly strong, with strength similar to that of the well-defined MYCN amplification in tumor cells. Soluble CD155 has been reported to be significantly increased in hepatocellular carcinoma (70) and in patients with neuroblastoma in our study (soluble CD112 displayed comparable levels in serum from patients with neuroblastoma and HC as shown in Supplementary Fig. S3E). However, we did not identify a difference in soluble CD155 levels between patients with different risk levels (Supplementary Fig. S3J). This is in contrast to the strong association among tumor cell membrane CD155 levels, tumor risk, and patient survival. This highlights the importance of our finding that downregulation of DNAM-1 expression in γδT cells requires direct tumor cell contact.

We found that DNAM-1 reduction induced by CD155 coexisted with γδT-cell functional exhaustion. Considering that CD155 was more prone to bind to TIGIT (KD = 1–3 nmol/L) than to DNAM-1 (KD = 115 nmol/L; ref. 71), it was possible that CD155 triggered an immune inhibitory signal through TIGIT in neuroblastoma γδT cells, similar to the suppressive effect on NK cells in other tumors (72). Upon rhCD155 protein treatment or coculture with neuroblastoma cell lines, the expression of TIGIT in γδT cells was downregulated via internalization along with DNAM-1 degradation. In contrast, upon coculture with primary neuroblastoma tumor cells or treatment with tumor tissue culture supernatant, upregulation of TIGIT was induced in intratumoral γδT cells. Therefore, the alternative pathways were mediated by unknown soluble factors within the TME that were involved in TIGIT regulation. Nevertheless, in both scenarios of different TIGIT regulation, ligation of CD155 with DNAM-1 induced a strong degradation of DNAM-1. Simply adding anti-TIGIT antibody (α-TIGIT) could not salvage the dominant γδT-cell suppression by CD155. Additionally, tumor tissue culture supernatant treatment did not induce DNAM-1 degradation, suggesting the requirement for direct contact between tumor cells and γδT cells. The effects of CD96 and CD112R were least likely to play determining roles, as we found a very low level of CD112R expressed on γδT cells and no change in CD96 expression in response to the above treatments. Notably, TIGIT showed a similar expression pattern to that of DNAM-1 in neuroblastoma tumors from the GSE49711 dataset. Lower TIGIT expression in neuroblastoma tumors with high risk, MYCN amplification, and patient death, and TIGIT expression levels was positively correlated with survival rates (Supplementary Fig. S2E–S2G). This finding was contradictory to the common recognition that TIGIT is an immunoinhibitory receptor on T cells. Multiple intertwined and sophisticated receptor–ligand interactions and ever-changing intratumoral microecosystems could possibly reconcile this disagreement.

Previous animal experiments have revealed that the expression of DNAM-1 is one of the main activators that determine T-cell competence in interacting with tumors. DNAM-1 retains immune cell activation even in the presence of exhaustion markers, such as TIM3 (58). CD112 has generally been reported as an activating ligand for DNAM-1 (20). CD155 is also involved in antitumor immune responses (22, 73), although it induces cell adhesion, migration, and proliferation of tumor cells (74). However, according to recent reports, the function of CD155 is controversial. Increasing data indicate that CD155 exerts inhibitory effects on T and NK cells by modulating DNAM-1, and the loss of CD155 enhances the antitumor function of immune cells. Additionally, DNAM-1 internalization and degradation in CD8 T cells correlate with CD155 expression on tumor cells (25, 58, 69). Our study is consistent with the tumor cell heterogeneity in CD155 and CD112 expression being closely related to the DNAM-1 expression levels in γδT cells and their cytotoxicity. Our findings suggest that CD155 plays a dominant inhibitory role responsible for the downregulation of DNAM-1 after γδT-cell and tumor cell contact, whereas the expression of CD112 on tumor cells retains and promotes the functional activation of γδT cells without affecting DNAM-1 expression. Moreover, CD112 has been found to be an especially important target for cytotoxicity against acute myeloid leukemia compared with CD155 (75), consistent with the crucial role of CD112 in γδT-cell activation in our study. This demonstrates that the heterogeneity of tumor cells, defined by multiple ligands interacting with DNAM-1 on immune cells through divergent mechanisms, ultimately shapes the amplitude of the antitumor immune response. Similar to the existence of γδT cells with differential DNAM-1 levels, our bulk RNA-seq data from solid neuroblastoma tumor reveal a broad degree of variance in CD112 and CD155 expression. Although CD155 alone acts as a strong inhibitory force for the immune response and shows significant predictive power for disease outcomes, logically adding CD112 based on our functional analysis provides a more accurate reflection of tumor cell fitness. Our ROC analysis based on the public neuroblastoma tumor GEO database revealed that the CD112/CD155 ratio exhibited a strong correlation with patient mortality, high risk, and MYCN amplification, demonstrating superior AUC values compared with CD155 alone (Supplementary Fig. S10C). Based on our bulk RNA-seq and scRNA-seq data, the arbitrary separation of tumors with a CD112/CD155 ratio represents two distinct functional subgroups in terms of cell metabolic rate and differentiation status. Functional evidence shows that tumor cells bearing high CD112/CD155 expression ratios are more susceptible to γδT-cell killing than tumor cells with low ratios in coculture.

According to our spatial transcriptomic analysis, tumor cells at different intratumoral locations exhibit dichotomous patterns of CD112 and CD155 expression. Although the evenly distributed pan-tissue presence of CD112 may interact with DNAM-1 to maintain immune cell activation, subregional high CD155 expression predominantly deters the infiltration and activation of immune cells. This implies that CD112 could be a secondary factor modulating basic immune activation, which is dictated by the CD155 expression level in tumor cells. Therefore, the CD112/CD155 ratio calculated from the public RNA-seq database could represent an average expression of the tumor tissue, reflecting the overall immune suppression power of the tumor. Undoubtedly, more immune cell receptor and tumor cell ligand pairs are expected to be involved in regulating the magnitude of immune activity. Within our research scope, we provide an example of the interplay between immune cells and heterogeneous tumor cells via one of the important receptor–ligand pairs reshaping each other, leading to different fates of tumor cells and divergent immune functions.

In our study, we dissected the effects of CD112 and CD155 on γδT-cell activation and their relation to DNAM-1. However, our data did not provide sufficient resolution to determine whether these two tumor molecules compete for DNAM-1 binding or if one directly affects the expression of the other. Precise work with high-resolution microscopy techniques, such as photoactivated localization microscopy (PALM), direct stochastic optical reconstruction microscopy (dSTORM) or stimulated emission depletion (STED), should be used in the future to demonstrate the expression of CD112 and CD155 at the single–tumor cell level. Based on our data, a lower quantity of more immunosuppressive γδT cells was associated with low-ratio tumor cells. Unfortunately, our spatial RNA-seq provided insufficient resolution to discern the exact sublocales of γδT cells and their juxtaposition with different tumor cells. Combining physical separation techniques, such as laser-capture microdissection, with scRNA-seq may provide more detailed information on the landscape of γδT cells and tumor cells within a single tumor.

Our spatial RNA-seq and IHC results clearly delineated the regions of high CD155 expression in a single tumor tissue sample. As indicated by our scRNA-seq, these CD112/CD155 low-ratio tumor cells were in a poorly differentiated state with significantly lower neuronal gene expression. Further studies should be conducted to determine whether these high CD155-expressing cells in particular tumor regions are derived from clonal expansion or are heterogeneous derivatives of the same phenotype.

X. Wang reports grants from the National Natural Science Foundation of China during the conduct of the study. X. Ni reports grants from the National Natural Science Foundation of China during the conduct of the study. J. Gui reports grants from the National Natural Science Foundation of China during the conduct of the study. No disclosures were reported by the other authors.

X. Wang: Conceptualization, data curation, software, formal analysis, funding acquisition, investigation, visualization, writing–original draft, writing–review and editing. H. Wang: Validation, investigation. Z. Lu: Validation, investigation, visualization. X. Liu: Formal analysis, methodology. W. Chai: Software, validation, methodology. W. Wang: Validation. J. Feng: Resources, formal analysis. S. Yang: Resources, formal analysis. W. Yang: Resources, formal analysis. H. Cheng: Resources. C. Chen: Resources, formal analysis. S. Zhang: Validation. N. Sun: Resources, formal analysis. Q. Liu: Resources. Q. Li: Resources. W. Song: Resources. F. Jin: Resources. Q. Zeng: Resources. S. Wang: Resources. Y. Su: Resources, validation. H. Wang: Resources, formal analysis. X. Ni: Resources, supervision, funding acquisition, writing–review and editing. J. Gui: Conceptualization, data curation, supervision, funding acquisition, project administration, writing–review and editing.

This work was supported by grants from the National Natural Science Foundation of China (numbers 82002637/82173084/82293660/82293665), Beijing Municipal Public Welfare Development and Reform Pilot Project for Medical Research Institutes (JYY2021-9), and Funding for Reform and Development of Beijing Municipal Health Commission.

Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

1.
Newman
EA
,
Abdessalam
S
,
Aldrink
JH
,
Austin
M
,
Heaton
TE
,
Bruny
J
, et al
.
Update on neuroblastoma
.
J Pediatr Surg
2019
;
54
:
383
9
.
2.
Ni
X
,
Li
Z
,
Li
X
,
Zhang
X
,
Bai
G
,
Liu
Y
, et al
.
Socioeconomic inequalities in cancer incidence and access to health services among children and adolescents in China: a cross-sectional study
.
Lancet
2022
;
400
:
1020
32
.
3.
Lundberg
KI
,
Treis
D
,
Johnsen
JI
.
Neuroblastoma heterogeneity, plasticity, and emerging therapies
.
Curr Oncol Rep
2022
;
24
:
1053
62
.
4.
Anderson
J
,
Majzner
RG
,
Sondel
PM
.
Immunotherapy of neuroblastoma: facts and hopes
.
Clin Cancer Res
2022
;
28
:
3196
206
.
5.
Yamaguchi
H
,
Hsu
JM
,
Yang
WH
,
Hung
MC
.
Mechanisms regulating PD-L1 expression in cancers and associated opportunities for novel small-molecule therapeutics
.
Nat Rev Clin Oncol
2022
;
19
:
287
305
.
6.
Wang
J
,
Sanmamed
MF
,
Datar
I
,
Su
TT
,
Ji
L
,
Sun
J
, et al
.
Fibrinogen-like protein 1 is a major immune inhibitory ligand of LAG-3
.
Cell
2019
;
176
:
334
47.e12
.
7.
Zhou
L
,
Ruan
M
,
Liu
Y
,
Zhu
Y
,
Fu
D
,
Wu
K
, et al
.
B7H4 expression in tumor cells impairs CD8 T cell responses and tumor immunity
.
Cancer Immunol Immunother
2020
;
69
:
163
74
.
8.
Molfetta
R
,
Quatrini
L
,
Capuano
C
,
Gasparrini
F
,
Zitti
B
,
Zingoni
A
, et al
.
c-Cbl regulates MICA- but not ULBP2-induced NKG2D down-modulation in human NK cells
.
Eur J Immunol
2014
;
44
:
2761
70
.
9.
Patel
AP
,
Tirosh
I
,
Trombetta
JJ
,
Shalek
AK
,
Gillespie
SM
,
Wakimoto
H
, et al
.
Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma
.
Science
2014
;
344
:
1396
401
.
10.
Tang
C
,
Xie
AX
,
Liu
EM
,
Kuo
F
,
Kim
M
,
DiNatale
RG
, et al
.
Immunometabolic coevolution defines unique microenvironmental niches in ccRCC
.
Cell Metab
2023
;
35
:
1424
40.e5
.
11.
Maestri
E
,
Kedei
N
,
Khatib
S
,
Forgues
M
,
Ylaya
K
,
Hewitt
SM
, et al
.
Spatial proximity of tumor-immune interactions predicts patient outcome in hepatocellular carcinoma
.
Hepatology
2024
;
79
:
768
79
.
12.
Shibuya
A
,
Campbell
D
,
Hannum
C
,
Yssel
H
,
Franz-Bacon
K
,
McClanahan
T
, et al
.
DNAM-1, a novel adhesion molecule involved in the cytolytic function of T lymphocytes
.
Immunity
1996
;
4
:
573
81
.
13.
Wang
X
,
Mou
W
,
Han
W
,
Xi
Y
,
Chen
X
,
Zhang
H
, et al
.
Diminished cytolytic activity of γδ T cells with reduced DNAM-1 expression in neuroblastoma patients
.
Clin Immunol
2019
;
203
:
63
71
.
14.
Zhang
H
,
Chai
W
,
Yang
W
,
Han
W
,
Mou
W
,
Xi
Y
, et al
.
The increased IL-17-producing γδT cells promote tumor cell proliferation and migration in neuroblastoma
.
Clin Immunol
2020
;
211
:
108343
.
15.
Chauvin
JM
,
Pagliano
O
,
Fourcade
J
,
Sun
Z
,
Wang
H
,
Sander
C
, et al
.
TIGIT and PD-1 impair tumor antigen-specific CD8+ T cells in melanoma patients
.
J Clin Invest
2015
;
125
:
2046
58
.
16.
Sanchez-Correa
B
,
Gayoso
I
,
Bergua
JM
,
Casado
JG
,
Morgado
S
,
Solana
R
, et al
.
Decreased expression of DNAM-1 on NK cells from acute myeloid leukemia patients
.
Immunol Cell Biol
2012
;
90
:
109
15
.
17.
Takahashi
N
,
Sugaya
M
,
Suga
H
,
Oka
T
,
Kawaguchi
M
,
Miyagaki
T
, et al
.
Increased soluble CD226 in sera of patients with cutaneous T-cell lymphoma mediates cytotoxic activity against tumor cells via CD155
.
J Invest Dermatol
2017
;
137
:
1766
73
.
18.
Conner
M
,
Hance
KW
,
Yadavilli
S
,
Smothers
J
,
Waight
JD
.
Emergence of the CD226 axis in cancer immunotherapy
.
Front Immunol
2022
;
13
:
914406
.
19.
Bottino
C
,
Castriconi
R
,
Pende
D
,
Rivera
P
,
Nanni
M
,
Carnemolla
B
, et al
.
Identification of PVR (CD155) and nectin-2 (CD112) as cell surface ligands for the human DNAM-1 (CD226) activating molecule
.
J Exp Med
2003
;
198
:
557
67
.
20.
El-Sherbiny
YM
,
Meade
JL
,
Holmes
TD
,
McGonagle
D
,
Mackie
SL
,
Morgan
AW
, et al
.
The requirement for DNAM-1, NKG2D, and NKp46 in the natural killer cell-mediated killing of myeloma cells
.
Cancer Res
2007
;
67
:
8444
9
.
21.
Focaccetti
C
,
Benvenuto
M
,
Pighi
C
,
Vitelli
A
,
Napolitano
F
,
Cotugno
N
, et al
.
DNAM-1-chimeric receptor-engineered NK cells, combined with nutlin-3a, more effectively fight neuroblastoma cells in vitro: a proof-of-concept study
.
Front Immunol
2022
;
13
:
886319
.
22.
Chan
CJ
,
Andrews
DM
,
McLaughlin
NM
,
Yagita
H
,
Gilfillan
S
,
Colonna
M
, et al
.
DNAM-1/CD155 interactions promote cytokine and NK cell-mediated suppression of poorly immunogenic melanoma metastases
.
J Immunol
2010
;
184
:
902
11
.
23.
Molfetta
R
,
Milito
ND
,
Zitti
B
,
Lecce
M
,
Fionda
C
,
Cippitelli
M
, et al
.
The ubiquitin-proteasome pathway regulates nectin2/CD112 expression and impairs NK cell recognition and killing
.
Eur J Immunol
2019
;
49
:
873
83
.
24.
Du
X
,
de Almeida
P
,
Manieri
N
,
de Almeida Nagata
D
,
Wu
TD
,
Harden Bowles
K
, et al
.
CD226 regulates natural killer cell antitumor responses via phosphorylation-mediated inactivation of transcription factor FOXO1
.
Proc Natl Acad Sci U S A
2018
;
115
:
E11731
40
.
25.
Li
XY
,
Das
I
,
Lepletier
A
,
Addala
V
,
Bald
T
,
Stannard
K
, et al
.
CD155 loss enhances tumor suppression via combined host and tumor-intrinsic mechanisms
.
J Clin Invest
2018
;
128
:
2613
25
.
26.
Okumura
G
,
Iguchi-Manaka
A
,
Murata
R
,
Yamashita-Kanemaru
Y
,
Shibuya
A
,
Shibuya
K
.
Tumor-derived soluble CD155 inhibits DNAM-1-mediated antitumor activity of natural killer cells
.
J Exp Med
2020
;
217
:
1
.
27.
Casado
JG
,
Pawelec
G
,
Morgado
S
,
Sanchez-Correa
B
,
Delgado
E
,
Gayoso
I
, et al
.
Expression of adhesion molecules and ligands for activating and costimulatory receptors involved in cell-mediated cytotoxicity in a large panel of human melanoma cell lines
.
Cancer Immunol Immunother
2009
;
58
:
1517
26
.
28.
Patel
M
,
Vlahava
VM
,
Forbes
SK
,
Fielding
CA
,
Stanton
RJ
,
Wang
E
.
HCMV-encoded NK modulators: lessons from in vitro and in vivo genetic variation
.
Front Immunol
2018
;
9
:
2214
.
29.
Prod’Homme
V
,
Sugrue
DM
,
Stanton
RJ
,
Nomoto
A
,
Davies
J
,
Rickards
CR
, et al
.
Human cytomegalovirus UL141 promotes efficient downregulation of the natural killer cell activating ligand CD112
.
J Gen Virol
2010
;
91
:
2034
9
.
30.
Nagumo
Y
,
Iguchi-Manaka
A
,
Yamashita-Kanemaru
Y
,
Abe
F
,
Bernhardt
G
,
Shibuya
A
, et al
.
Increased CD112 expression in methylcholanthrene-induced tumors in CD155-deficient mice
.
PLoS One
2014
;
9
:
e112415
.
31.
Siegers
GM
,
Dutta
I
,
Lai
R
,
Postovit
LM
.
Functional plasticity of gamma delta T cells and breast tumor targets in hypoxia
.
Front Immunol
2018
;
9
:
1367
.
32.
Harmon
C
,
Zaborowski
A
,
Moore
H
,
St Louis
P
,
Slattery
K
,
Duquette
D
, et al
.
γδ T cell dichotomy with opposing cytotoxic and wound healing functions in human solid tumors
.
Nat Cancer
2023
;
4
:
1122
37
.
33.
Zhang
W
,
Yu
Y
,
Hertwig
F
,
Thierry-Mieg
J
,
Zhang
W
,
Thierry-Mieg
D
, et al
.
Comparison of RNA-seq and microarray-based models for clinical endpoint prediction
.
Genome Biol
2015
;
16
:
133
.
34.
Scaruffi
P
,
Morandi
F
,
Gallo
F
,
Stigliani
S
,
Parodi
S
,
Moretti
S
, et al
.
Bone marrow of neuroblastoma patients shows downregulation of CXCL12 expression and presence of IFN signature
.
Pediatr Blood Cancer
2012
;
59
:
44
51
.
35.
Dessau
RB
,
Pipper
CB
.
[“R”—project for statistical computing]
.
Ugeskr Laeger
2008
;
170
:
328
30
.
36.
Ritchie
ME
,
Phipson
B
,
Wu
D
,
Hu
Y
,
Law
CW
,
Shi
W
, et al
.
limma powers differential expression analyses for RNA-sequencing and microarray studies
.
Nucleic Acids Res
2015
;
43
:
e47
.
37.
Hubbard
N
,
Hagin
D
,
Sommer
K
,
Song
Y
,
Khan
I
,
Clough
C
, et al
.
Targeted gene editing restores regulated CD40L function in X-linked hyper-IgM syndrome
.
Blood
2016
;
127
:
2513
22
.
38.
Wu
T
,
Hu
E
,
Xu
S
,
Chen
M
,
Guo
P
,
Dai
Z
, et al
.
clusterProfiler 4.0: a universal enrichment tool for interpreting omics data
.
Innovation (Camb)
2021
;
2
:
100141
.
39.
Subramanian
A
,
Tamayo
P
,
Mootha
VK
,
Mukherjee
S
,
Ebert
BL
,
Gillette
MA
, et al
.
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
.
Proc Natl Acad Sci U S A
2005
;
102
:
15545
50
.
40.
Dobin
A
,
Davis
CA
,
Schlesinger
F
,
Drenkow
J
,
Zaleski
C
,
Jha
S
, et al
.
STAR: ultrafast universal RNA-seq aligner
.
Bioinformatics
2013
;
29
:
15
21
.
41.
Butler
A
,
Hoffman
P
,
Smibert
P
,
Papalexi
E
,
Satija
R
.
Integrating single-cell transcriptomic data across different conditions, technologies, and species
.
Nat Biotechnol
2018
;
36
:
411
20
.
42.
Jansky
S
,
Sharma
AK
,
Körber
V
,
Quintero
A
,
Toprak
UH
,
Wecht
EM
, et al
.
Single-cell transcriptomic analyses provide insights into the developmental origins of neuroblastoma
.
Nat Genet
2021
;
53
:
683
93
.
43.
Xue
R
,
Zhang
Q
,
Cao
Q
,
Kong
R
,
Xiang
X
,
Liu
H
, et al
.
Liver tumour immune microenvironment subtypes and neutrophil heterogeneity
.
Nature
2022
;
612
:
141
7
.
44.
Combes
AN
,
Zappia
L
,
Er
PX
,
Oshlack
A
,
Little
MH
.
Single-cell analysis reveals congruence between kidney organoids and human fetal kidney
.
Genome Med
2019
;
11
:
3
.
45.
Mazzurana
L
,
Czarnewski
P
,
Jonsson
V
,
Wigge
L
,
Ringnér
M
,
Williams
TC
, et al
.
Tissue-specific transcriptional imprinting and heterogeneity in human innate lymphoid cells revealed by full-length single-cell RNA-sequencing
.
Cell Res
2021
;
31
:
554
68
.
46.
Tirosh
I
,
Izar
B
,
Prakadan
SM
,
Wadsworth
MH
2nd
,
Treacy
D
,
Trombetta
JJ
, et al
.
Dissecting the multicellular ecosystem of metastatic melanoma by single-cell RNA-seq
.
Science
2016
;
352
:
189
96
.
47.
Young
MD
,
Mitchell
TJ
,
Vieira Braga
FA
,
Tran
M
,
Stewart
BJ
,
Ferdinand
JR
, et al
.
Single-cell transcriptomes from human kidneys reveal the cellular identity of renal tumors
.
Science
2018
;
361
:
594
9
.
48.
Bedoya-Reina
OC
,
Li
W
,
Arceo
M
,
Plescher
M
,
Bullova
P
,
Pui
H
, et al
.
Single-nuclei transcriptomes from human adrenal gland reveal distinct cellular identities of low and high-risk neuroblastoma tumors
.
Nat Commun
2021
;
12
:
5309
.
49.
Luo
Z
,
He
J
,
Chen
Y
,
He
J
,
Gong
T
,
Tang
F
, et al
.
Air flow-assisted ionization imaging mass spectrometry method for easy whole-body molecular imaging under ambient conditions
.
Anal Chem
2013
;
85
:
2977
82
.
50.
Cheng
A
,
Xu
Q
,
Li
B
,
Zhang
L
,
Wang
H
,
Liu
C
, et al
.
The enhanced energy metabolism in the tumor margin mediated by RRAD promotes the progression of oral squamous cell carcinoma
.
Cell Death Dis
2024
;
15
:
376
.
51.
Race
AM
,
Styles
IB
,
Bunch
J
.
Inclusive sharing of mass spectrometry imaging data requires a converter for all
.
J Proteomics
2012
;
75
:
5111
2
.
52.
Bemis
KD
,
Harry
A
,
Eberlin
LS
,
Ferreira
C
,
van de Ven
SM
,
Mallick
P
, et al
.
Cardinal: an R package for statistical analysis of mass spectrometry-based imaging experiments
.
Bioinformatics
2015
;
31
:
2418
20
.
53.
Wang
G
,
Heijs
B
,
Kostidis
S
,
Mahfouz
A
,
Rietjens
R
,
Bijkerk
R
, et al
.
Analyzing cell-type-specific dynamics of metabolism in kidney repair
.
Nat Metab
2022
;
4
:
1109
18
.
54.
Palmer
A
,
Phapale
P
,
Chernyavsky
I
,
Lavigne
R
,
Fay
D
,
Tarasov
A
, et al
.
FDR-controlled metabolite annotation for high-resolution imaging mass spectrometry
.
Nat Methods
2017
;
14
:
57
60
.
55.
Wagner
A
,
Wang
C
,
Fessler
J
,
DeTomaso
D
,
Avila-Pacheco
J
,
Kaminski
J
, et al
.
Metabolic modeling of single Th17 cells reveals regulators of autoimmunity
.
Cell
2021
;
184
:
4168
85.e21
.
56.
Soriani
A
,
Zingoni
A
,
Cerboni
C
,
Iannitto
ML
,
Ricciardi
MR
,
Di Gialleonardo
V
, et al
.
ATM-ATR-dependent up-regulation of DNAM-1 and NKG2D ligands on multiple myeloma cells by therapeutic agents results in enhanced NK-cell susceptibility and is associated with a senescent phenotype
.
Blood
2009
;
113
:
3503
11
.
57.
Tran
HC
,
Marachelian
A
,
Venkatramani
R
,
Jubran
RF
,
Mascarenhas
L
.
Oxaliplatin and doxorubicin for relapsed or refractory high-risk neuroblastoma
.
Pediatr Hematol Oncol
2015
;
32
:
26
31
.
58.
Braun
M
,
Aguilera
AR
,
Sundarrajan
A
,
Corvino
D
,
Stannard
K
,
Krumeich
S
, et al
.
CD155 on tumor cells drives resistance to immunotherapy by inducing the degradation of the activating receptor CD226 in CD8+ T cells
.
Immunity
2020
;
53
:
805
23.e15
.
59.
Zhang
Z
,
Wu
N
,
Lu
Y
,
Davidson
D
,
Colonna
M
,
Veillette
A
.
DNAM-1 controls NK cell activation via an ITT-like motif
.
J Exp Med
2015
;
212
:
2165
82
.
60.
O’Leary
CE
,
Lewis
EL
,
Oliver
PM
.
Ubiquitylation as a rheostat for TCR signaling: from targeted approaches toward global profiling
.
Front Immunol
2015
;
6
:
618
.
61.
Li
JY
,
Zhao
Y
,
Gong
S
,
Wang
MM
,
Liu
X
,
He
QM
, et al
.
TRIM21 inhibits irradiation-induced mitochondrial DNA release and impairs antitumour immunity in nasopharyngeal carcinoma tumour models
.
Nat Commun
2023
;
14
:
865
.
62.
Chen
X
,
Cao
M
,
Wang
P
,
Chu
S
,
Li
M
,
Hou
P
, et al
.
The emerging roles of TRIM21 in coordinating cancer metabolism, immunity and cancer treatment
.
Front Immunol
2022
;
13
:
968755
.
63.
Gröbner
SN
,
Worst
BC
,
Weischenfeldt
J
,
Buchhalter
I
,
Kleinheinz
K
,
Rudneva
VA
, et al
.
The landscape of genomic alterations across childhood cancers
.
Nature
2018
;
555
:
321
7
.
64.
Prigione
I
,
Corrias
MV
,
Airoldi
I
,
Raffaghello
L
,
Morandi
F
,
Bocca
P
, et al
.
Immunogenicity of human neuroblastoma
.
Ann N Y Acad Sci
2004
;
1028
:
69
80
.
65.
Silva-Santos
B
,
Serre
K
,
Norell
H
.
γδ T cells in cancer
.
Nat Rev Immunol
2015
;
15
:
683
91
.
66.
Sebestyen
Z
,
Prinz
I
,
Dechanet-Merville
J
,
Silva-Santos
B
,
Kuball
J
.
Translating gammadelta (γδ) T cells and their receptors into cancer cell therapies
.
Nat Rev Drug Discov
2020
;
19
:
169
84
.
67.
Song
Y
,
Liu
Y
,
Teo
HY
,
Liu
H
.
Targeting cytokine signals to enhance γδT cell-based cancer immunotherapy
.
Front Immunol
2022
;
13
:
914839
.
68.
Gentles
AJ
,
Newman
AM
,
Liu
CL
,
Bratman
SV
,
Feng
W
,
Kim
D
, et al
.
The prognostic landscape of genes and infiltrating immune cells across human cancers
.
Nat Med
2015
;
21
:
938
45
.
69.
Briukhovetska
D
,
Suarez-Gosalvez
J
,
Voigt
C
,
Markota
A
,
Giannou
AD
,
Schübel
M
, et al
.
T cell-derived interleukin-22 drives the expression of CD155 by cancer cells to suppress NK cell function and promote metastasis
.
Immunity
2023
;
56
:
143
61
.
70.
Jin
AL
,
Yang
YH
,
Su
X
,
Yang
WJ
,
Liu
T
,
Chen
W
, et al
.
High serum soluble CD155 level predicts poor prognosis and correlates with an immunosuppressive tumor microenvironment in hepatocellular carcinoma
.
J Clin Lab Anal
2022
;
36
:
e24259
.
71.
Manieri
NA
,
Chiang
EY
,
Grogan
JL
.
TIGIT: a key inhibitor of the cancer immunity cycle
.
Trends Immunol
2017
;
38
:
20
8
.
72.
Sanchez-Correa
B
,
Valhondo
I
,
Hassouneh
F
,
Lopez-Sejas
N
,
Pera
A
,
Bergua
JM
, et al
.
DNAM-1 and the TIGIT/PVRIG/TACTILE axis: novel immune checkpoints for natural killer cell-based cancer immunotherapy
.
Cancers (Basel)
2019
;
11
:
877
.
73.
Lupo
KB
,
Matosevic
S
.
CD155 immunoregulation as a target for natural killer cell immunotherapy in glioblastoma
.
J Hematol Oncol
2020
;
13
:
76
.
74.
Molfetta
R
,
Zitti
B
,
Lecce
M
,
Milito
ND
,
Stabile
H
,
Fionda
C
, et al
.
CD155: a multi-functional molecule in tumor progression
.
Int J Mol Sci
2020
;
21
:
922
.
75.
Kaito
Y
,
Sugimoto
E
,
Nakamura
F
,
Tsukune
Y
,
Sasaki
M
,
Yui
S
, et al
.
Immune checkpoint molecule DNAM-1/CD112 axis is a novel target for natural killer-cell therapy in acute myeloid leukemia
.
Haematologica
2024
;
109
:
1107
20
.
This open access article is distributed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license.

Supplementary data