NK cell isolation and expansion
Peripheral blood mononuclear cells (PBMC) were obtained from healthy donors (HD) buffy coat from the transfusion center of IRCCS Bambino Gesù Children’s Hospital, Rome (Approval of Ethical Committee prot. N° 729) and isolated after density gradient centrifugation (Ficoll-Lympholyte, Cederlane). NK cells were purified using NK isolation kit II (Miltenyi Biotec) or Rosettsep 10 (StemCell) and shortly activated (72 h) with IL15 (150U/ml, Miltenyi Biotec), IL2 (50U/ml), and IL1β (2000U/ml, Miltenyi Biotec) in NK MACS medium (Miltenyi Biotec) supplemented with 5% human serum and 2 mM Glutamax. Purity was evaluated by flow cytometry; only >96% of purity was considered suitable for cultures. Cells were maintained in a humidified atmosphere containing 8% CO2 at 37 °C.
Generation of γ-retroviral vector and NK cell transduction
A γ-retroviral vector was constructed to encode a chimeric construct comprising PD1-NKG2D.4-1BB.2 A.NKG2A(scFv).2A.IL15. This tri-cistronic vector enables the expression of a transmembrane fusion receptor containing the PD-1 ectodomain, designed to bind PD-1 ligands, fused in-frame with a secretable scFv targeting NKG2A and the cytokine IL15.
To validate the contribution of each engineered component of our γ-retroviral vector, we generated three new tri-cistronic vectors by selectively replacing individual components while retaining the same backbone and expression configuration. We developed tri-cistronic vectors in which the NKG2A-scFv or IL15 transgenes, or both, were replaced with an irrelevant protein (IP), referred to as w/o-NKG2A, w/o-IL15, or w/o-NKG2A/IL15. IP for NKG2A-scFv was ΔCD19, while for IL15 was red fluorescent protein (RFP). These modifications preserved the overall architecture of the vectors while specifically removing NKG2A and/or IL15 from the system.
Retroviral supernatant was produced by transient transfection of HEK-293T cells as described previously,54,55 and viral titers were quantified using the Retro-X™ qRT-PCR Titration Kit (Takara Bio). Human NK cells were cultured in expansion medium for 3 days before transduction, then seeded into the RetroNectin-coated plates (Takara Bio, Japan) containing the spin-inoculated viral supernatant. The culture medium was replaced after 72 hours with NK MACS Medium plus 5% human serum and 2 mM Glutamax with IL15 (150U/ml, Miltenyi Biotec) and IL2 (50U/ml). Transduction efficiency was assessed by flow cytometry.
Cell lines and cultures
The NALM-18 (Childhood B acute lymphoblastic leukemia), Karpas (human lymphoma), HL-60 (acute promyelocytic leukemia), HLA-E+ LCL 721.221.G tumor cells, HEK-293T, and THP-1 (acute monocytic leukemia) cancer cell line were kindly provided by Dr Pende D. (IRCCS, Policlinico San Martino, Genoa, Italy) and cultured in RPMI 1640 (Euroclone, Milan, Italy) supplemented with 10% FBS (Euroclone), 1% penicillin/streptomycin (Euroclone), and 1% L-glutamine (Euroclone). SKNAS, SKNSH, IMR32, and SHSY5Y neuroblastoma cell lines (ATCC) were cultured in Dulbecco’s modified Eagle’s (DMEM) supplemented with 10% FBS, 1% penicillin/streptomycin, and 1% L-glutamine. D283-Med and D341-Med cells were obtained from the American Type Culture Collection (ATCC). Cell lines were grown in MEM supplemented with 20% of Fetal Bovine Serum (FBS), 1% of MEM Non-Essential Amino Acids (NEAA), 1% of Sodium pyruvate, and 1% of PenStrep. All cell lines were cultured in a humidified atmosphere with 5% CO2 at 37 °C. For the generation of eGFP-Firefly-Luciferase (FF/GFP) cell lines, the retroviral vector encoding eGFP-Firefly-Luciferase was used in selected experiments to label positive (PD-L1+) or negative (PD-L1-) tumor cells. Cells were maintained in a humidified atmosphere containing 5% CO2 at 37°C. All cell lines were routinely 15 tested for mycoplasma and surface expression of target antigens.
Protein purification and characterization
The anti-NKG2A scFv was purified from the conditioned media of HEK-293T mammalian cells expressing proteins using affinity columns and characterized. Particularly, the scFv was purified on a Pierce Protein L Plus Agarose column (Thermoscientific) according to the manufacturer’s instructions. scFv was dialyzed against phosphate buffer saline (PBS) overnight at +4 °C and sterile filtered using a Millex-GP 0.22 μm filter unit (Millipore). Subsequently, it was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using NuPAGE Novex Bis-Tris gel (Thermo Scientific) under reducing conditions.
Detection of IL15 by ELISA assay
HEK-293T cells, exPD1- and mock-NK cells were cultured for different time points and cellular concentration in media containing 10% FBS, 1% Pen/Strep, and 2 mM Glutamax. Levels of IL15were measured using IL15 human ELISA kits (R&D Systems, DY247-05). All procedures were performed as described in the manufacturer’s instructions.
Immunoblotting
For the evaluation of protein levels, the cell pellet was lysed in RIPA buffer supplemented with protease and phosphatase inhibitors, then incubated on ice for 20 minutes. Cell lysates were centrifuged at 15,000 g for 15 minutes to remove insoluble components. The supernatants were collected, and proteins were quantified using the Bradford assay (Bio-Rad). Cell lysate was heated at 85 °C for 2 min in 4X Bolt™ LDS Sample Buffer (Invitrogen) and loaded on precast Bolt 4–12% Bis-Tris Plus Gels (Invitrogen). After electrophoresis, proteins were transferred to a nitrocellulose membrane (Amersham, GE Healthcare). The membranes were blocked in 5% nonfat dry milk (Cell Signaling Technology) in PBST (PBS containing 0.05% Tween 20) for one hour at room temperature (RT) and then incubated overnight at 4 °C with the following primary antibodies: Anti-p-JNK (Cell Signaling Technology #mAb 9251S), Anti-p-cJun (Cell Signaling Technology #mAb 91952), Anti-p-NFkB (Cell Signaling Technology #mAb 3033S), and Anti-β-Actin (Sigma-Aldrich A5441). Following 1-hour incubation with HRP-conjugated secondary antibodies (Cell Signaling Technology goat anti-rabbit IgG #7074 and goat-anti-mouse IgG #7076S), the signal was detected using ECL Prime Western Blotting Detection Reagent (Amersham, GE Healthcare) on a Uvitec Cambridge Mini HD9.
Monoclonal antibodies and cytofluorimetric analysis
For cytofluorimetric analysis, cells were stained with surface antibodies in PBS containing 5% FCS for 20 min at 4 °C. Expression of cell surface molecules was determined by flow-cytometry using standard methodology. The following mAbs were used: CD3-APC (clone REA613 | SK7, Miltenyi Biotec), CD19-Fitc (clone LT19, Miltenyi Biotec), PD-1-PE ((clone PD1.3.1.3, Miltenyi Biotec), NKG2D-PE/DAZZLE (clone 1D11, Biolegend), DNAM1-PC7 (clone 11A8, Biolegend), NKp44-APC (clone P44-8, Biolegend), CD45-APC700 (clone H130, Biolegend), CD57-PB (clone HNK-1, Biolegend), CD56-PC7 (clone N901, Beckman Coulter), CD56-BV650 (clone NCAM16.2, BD), NKG2A-APC (Z199, Beckman Coulter), CD94-FITC (DX22,BioLegend), PD-L1-PECF594 (clone MIH1, BD), PD-L2-PEVio615 (clone REA985 | MIH18, Miltenyi Biotec), HLA-E-PE (clone 3D12, BD), Macs Marker Screen V2 (Miltenyi Biotec). For the proliferation assays, HD-NK cells were labeled with CellTrace™ CFSE Cell Proliferation Kit (Invitrogen, #C34554) following the manufacturer’s instructions and resuspended in RPMI1640 containing 10% FBS, 1% Pen/Strep, and 2 mM Glutamax. CFSE-HD-NK cells were co-cultured in transwell conditions with genetically modified NK cells. At day 7, HD-NK cells (bottom wells) were harvested and analyzed by flow cytometry. In selected co-culture experiments, an anti-IL15 mAb was added (Bio-Techne, #MAB247, 5ug/ml). Samples were analyzed on a Cytoflex S and Cytoflex LX (Beckman Coulter). Data were analyzed using Cytexpert (Beckman Coulter) and FlowJo 10 software (BD Biosciences). For each sample, we analyzed a minimum of 50,000 events.
Cytotoxic assay
The cytotoxic assay was performed as previously described56 using as a medium: RPMI1640 containing 10% FBS, 1% Pen/Strep, and 2 mM Glutamax. Briefly, target cells were either labeled with Green Cell Tracker (CMFDA; Invitrogen, Thermo Fisher Scientific) according to the manufacturer’s instructions or directly used target cells transduced with the eGFP-Firefly-Luciferase vector. Target cells were incubated with NK cells at different effector/target (E:T) ratios. After 4 hours (h) of incubation at 37°C, propidium iodide (PI, Sigma-Aldrich) was added to detect dead tumor cells. Cells were acquired with the Beckman-Coulter Cytoflex-S/LX flow cytometer, and live target cells were identified as CMFDA+ PI− or GFP+ PI−, whereas dead target cells were CMFDA+ PI+ or GFP+ PI+. The percentage of cell lysis was calculated as follows: % cell lysis = ((% of dead cells cultured with NK)−(% of spontaneous lysis)) / (100 − % of spontaneous lysis) × 100. For long-time (48 h) co-culture experiments, exPD1- and mock-NK cells were co-cultured at a 4:1, 1:1, or 1:4 E:T ratio with tumor cells. Following 2 days of incubation at 37°C, residual tumor cells and NK cells were evaluated by flow-cytometry analysis.
Seahorse extracellular flux analysis
Seahorse experiments were performed on isolated NK cells using the XF Cell Mito Stress kit (Seahorse Bioscience). OCR and ECAR were measured with XF96 Extracellular Flux Analyzers (Seahorse Bioscience). Briefly, NK cells were plated on poly-d-lysine-coated 96-well polystyrene Seahorse plates (50 000 NK cells/well), equilibrated for 1 hour at 37 °C, and assayed for OCR (pmol/min) and ECAR (mpH/min) in basal conditions and after addition of oligomycin (1 μM), carbonyl cyanide-4-phenylhydrazone (1.5 μM), and antimycin A/rotenone (1 μM/0.1 μM). All of the SeaHorse experiments were performed using the manufacturer’s recommended media (pH 7.4), without phenol red, to standardize the pH conditions in all samples.
Real-time tumor cell cytotoxicity assays
To discriminate tumor and NK cells in co-culture, SK-N-AS and IMR-32 cells were transduced with a nuclear-restricted near-IR fluorescent label following the manufacturer’s instructions (Sartorius). Target tumor cells were seeded in 96-well flat-bottom plates. After overnight incubation, Caspase -3/7 Green Dye (Sartorius) and effector cells were added to the medium. Tumor and NK cells were co-cultured at a 1:1 E: T ratio for 72 h in the Incucyte Live-Cell Analysis System. The assay plate scan will be scheduled every 4 hours. Tumor cells will be counted with the IncuCyte SX5 Live-Cell Analysis System every time scanned to monitor tumor growth.
3D human pancreatic PDO culture
Patient-derived organoids were kindly provided by Prof. Dr. MD. Maximilian Reichert (Technical University of Munich (TUM), Germany. PDO were growth in Matrigel (Corning) domes in 24-well plates in feeding media contained the following components: AdDMEM/F12 medium supplemented GlutaMax (1X, Life Technologies), B27 (1X, Life Technologies), Primocin (100 μg/mL, InvivoGen), N-acetyl-L-cysteine (1.25 mM, Sigma-Aldrich), Nicotinamide (10 mM, Sigma-Aldrich), recombinant human Wnt3a protein (100 ng/mL, R&D Systems), mNoggin (100 ng/mL, Preprotech), EGF (50 ng/mL, Life Technologies), Gastrin (10 nM, Sigma-Aldrich), FGF10 (100 ng/mL, Preprotech), and A83- 01 (0.5 μM, Tocris). For organoids passing, the feeding medium was aspirated, and cold PBS was added to the well to dissolve the Matrigel matrix. Then, the mixture of organoid, Matrigel, and PBS was transferred into a 15 ml tube and centrifuged. The organoid pellet was dissociated by enzymatic digestion for 5 to 10 minutes with TrypLE (Life Technologies) at 37 °C. After dissociation, the organoids were washed with cold PBS, mixed with Matrigel, and plated in a prewarmed 24-well plate.
Spheroids of MB cells
To assess the impact of tumor architecture, GFP+D283-Med and D341-Med cells were seeded in 96 96-well ULA plate (cat#7007, Corning Incorporated) at very low cell density in their culture media supplemented with 0.25% methylcellulose (cat#M7027, Sigma-Aldrich), as previously described.57 After 72 h, spheroids were co-cultured with NK cells at different E:T ratios for additional 24 or 48 h. Spheroid growth and NK cell-mediated cytotoxicity were checked at various time points using the Celigo Image Cytometer (Nexcelom Bioscience), a microplate-based multichannel automated imaging system, to evaluate changes in area and fluorescence intensity.58 NK cell cytotoxic potential against treated spheroids was assessed using a Cytoflex LX flow cytometer (Beckman Coulter) following PI staining.
3D cytotoxic assay
PDOs were grown in a Matrigel matrix for 5 days, and then the NK cells were added to the medium. NK cells and PDO were co-cultured for 72 hours in the IncuCyte SX5 Live-Cell Analysis System. The assay plate scan was scheduled every 24 hours. The ability of NK cells to migrate to PDO and kill them was evaluated by measuring the PDO area at different time points using the ImageJ software.
In vivo experiments
To investigate the in vivo antitumor activity of exPD1-NK cells, leukemia and neuroblastoma mice tumor models were used. NALM-18-FF/GFP cells were intravenously (i.v.) injected (0.25 x 106 cells), in 6–8-week-old female NSG mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ; from Charles River). In a solid tumor setting, immunodeficient NSG mice were subcutaneously or orthotopically engrafted with SK-N-AS FF/GFP neuroblastoma. For the orthotopic MB models, D283-Med FF/GFP (15×104) cells were injected in the cerebellum of NSG mice (intracranial) using stereotaxic, as previously described.59 After tumor engraftment, the mice received an i.v. injection of exPD1-NK cells or mock-NK cells (7×106/mouse) harvested, washed, and resuspended in PBS 1X. Untreated mice served only as negative controls and received an i.v. injection of an equal volume of PBS 1X. Tumor growth was evaluated using the IVIS imaging system (PerkinElmer, USA) or the caliper. The intensity of the signal was measured every week as total photons/sec/cm2/sr (p/s/cm2/sr), as previously described.60 Mice were maintained in the animal facility at Plaisant Castel Romano (Rome, Italy). All in vivo experiments complied with the ethical international, EU, and national requirements and were approved by an ethical committee (Italian Health Ministry N°88/2016-PR and 805/2024-PR). Mice were matched based on the tumor signal for control and treatment groups before infusion of mock or exPD1-NK cells. To compare the growth of tumors over time, bioluminescence signal intensity or tumor size measured with a caliper was collected blindly. Bioluminescence signal intensity was log-transformed and then compared using a two-sample t-test. In addition to tumor growth kinetics, in order to evaluate the toxicity of treatments, animal welfare was carefully assessed on a daily basis by trained personnel using standardized clinical scoring criteria, including body weight loss, posture, mobility, coat condition, food and water intake, and signs of pain or distress. Peripheral blood was collected from tumor-bearing mice treated or not with mock- or exPD1-NK cells at different time points. Blood samples were processed to obtain serum, and biochemical analyses were performed to assess: aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (ALP), albuminemia, total bilirubin, creatinine, and lactate dehydrogenase (LDH) by Plaisant Castel Romano (Rome, Italy). At the end of the experiment, spleens were harvested, fixed, and processed for histological evaluation. Tissue sections were prepared from fixed tissue and subjected to hematoxylin and eosin (H&E) staining. Histological analysis was performed independently by two operators. The circulating human exPD1-NK cells were evaluated at the end of the experiments in selected organs.
Proteomic analysis
Protein quantification was performed using mass spectrometry with a data-independent acquisition (DIA) approach. Proteins detected in at least 50% of samples in at least one experimental group were retained for downstream analysis. Normality of protein expression was assessed with the Shapiro-Wilk test. Proteins with a normal distribution (p-value > 0.05) were compared between groups using a parametric test such as limma,61 which is a moderated Analysis of Variance (ANOVA) test, followed by Benjamini-Hochberg correction for multiple testing. Conversely, non-normally distributed proteins were analyzed with a non-parametric Wilcoxon-Mann-Whitney test. Pathway enrichment was performed on proteins with unadjusted p-values < 0.05 using the Gene Ontology category “biological processes” to provide a comprehensive overview of group-specific biological differences. Enrichment results were corrected for multiple testing using the Benjamini-Hochberg procedure. For each pathway, the Gene Ratio (number of proteins mapped to the pathway / total input proteins), Background Ratio (number of proteins in the pathway / total database proteins), and Count of mapped proteins were reported. Finally, Gene Set Enrichment Analysis (GSEA)62,63 was calculated by the GSEApy Python package64 to assess whether specific pathways were consistently up- or down-regulated between sample groups, providing insight into coordinated changes in protein expression.
Data acquisition from TCGA and TARGET
Gene expression and clinical data for 33 adult cancer types were retrieved from The Cancer Genome Atlas (TCGA) using the R/Bioconductor package TCGAbiolinks (v2.26.3). Expression data were downloaded from the Transcriptome Profiling category, specifically selecting the Gene Expression Quantification data type with the STAR-Counts workflow. The following TCGA projects were included: TCGA-LGG, TCGA-THYM, TCGA-KIRC, TCGA-PAAD, TCGA-SKCM, TCGA-TGCT, TCGA-BRCA, TCGA-SARC, TCGA-PCPG, TCGA-UVM, TCGA-ACC, TCGA-HNSC, TCGA-COAD, TCGA-LAML, TCGA-UCS, TCGA-GBM, TCGA-LIHC, TCGA-OV, TCGA-KIRP, TCGA-KICH, TCGA-READ, TCGA-THCA, TCGA-LUAD, TCGA-MESO, TCGA-CHOL, TCGA-PRAD, TCGA-STAD, TCGA-LUSC, TCGA-DLBC, TCGA-CESC, TCGA-ESCA, TCGA-BLCA, and TCGA-UCEC. Clinical data were retrieved independently using analogous procedures, querying the Clinical data category with the Clinical Supplement data type. In addition, pediatric tumor data were obtained from the TARGET (Therapeutically Applicable Research to Generate Effective Treatments) database using TCGAbiolinks. The following pediatric cancer projects were included: TARGET-NBL (Neuroblastoma), TARGET-RT (Rhabdoid Tumor), TARGET-WT (Wilms Tumor), TARGET-CCSK (Clear Cell Sarcoma of the Kidney), TARGET-ALL-P3, TARGET-ALL-P2, TARGET-ALL-P1 (Acute Lymphoblastic Leukemia phases 3, 2, and 1, respectively), TARGET-AML (Acute Myeloid Leukemia), and TARGET-OS (Osteosarcoma). Gene expression values in Transcripts Per Million (TPM) were extracted from transcript quantification files available for each TCGA/TARGET sample.
Gene signatures and scoring
We defined multiple custom gene signatures to capture distinct immunological and tumor-related programs from TCGA/TARGET RNA-seq data. Signature scores were calculated as the mean expression (TPM values) of all genes in each signature detected in a sample; if fewer than three genes were present, the score was set to NA to ensure robustness, otherwise the arithmetic mean was computed, providing a summary metric reflecting the overall activity of the biological program represented by each signature. An immune infiltration signature of 7 genes (CD8A, CD8B, CXCL9, CXCL10, IFNG, GZMB, PRF1) was used to estimate general immune cell infiltration and stratify tumor samples into “cold” and “hot” based on the first and the last quantiles. Tumor samples with scores at or below the 20th percentile were classified as “cold” (low immune infiltration). Conversely, those tumor samples at or above the 80th percentile were classified as “hot” (high immune infiltration). A more specific NK cell infiltration signature of 20 genes (KLRK1, NCAM1, KLRD1, KLRC1, KLRC2, KLRB1, GZMB, GZMA, GNLY, PRF1, FCGR3A, NKG7, XCL1, XCL2, TYROBP, IL2RB, ZAP70, SH2D1B, CD160, CD244) was used to reflect natural killer cell presence in the tumor. Finally, an immunosuppression signature of 18 genes (PDCD1, TIGIT, HAVCR2, CD96, KLRC1, LAG3, CISH, SOCS1, SOCS3, IL10RA, TGFBR2, EOMES, FOXP3, CEACAM1, BTLA, ZFP36, PRDM1, IKZF2) was employed to represent checkpoint and inhibitory immune regulation within the TME. The correlations between these signatures and PD-L1/HLA-E expression were assessed by Pearson correlation analysis.
Survival analysis
To assess the prognostic relevance of the selected ligand genes across cancers, we performed a Kaplan-Meier analysis. Clinical metadata, including vital status and time to death or last follow-up, were extracted from TCGA XML files and matched to gene expression quantifications (TPM values) obtained from RNA-seq data. For each tumor type, patients were stratified into “High” and “Low” expression groups based on the top and bottom quintiles (80th and 20th quantiles) of tumor ligand expression. Kaplan-Meier survival curves were generated for each tumor-ligand pair, and statistical significance was evaluated using the log-rank test. Only tumor-ligand combinations with sufficient sample size (n ≥ 10 per group) were considered. In parallel, we fitted univariate Cox proportional hazards models to quantify the hazard ratio (HR) associated with high versus low expression of each ligand across tumors. The log10-transformed HRs were visualized in a heatmap to summarize survival trends, with significance (p-value < 0.05) annotated directly on the plot. This approach enabled cross-tumor comparison of ligand prognostic impact, highlighting context-specific survival associations.
