Mice

BALB/c nude mice aged 6–8 weeks were purchased from GemPharmatech. All mice were maintained in specific pathogen-free condition in animal center of East China Normal University. All mouse experiments were performed in accordance with national guidelines for the humane treatment of animals (National animal treatment rules). The protocols were approved by East China Normal University Ethics Committee (ethical approval number m20210602).

Cell lines and Culture condition

Prostate cancer cell lines PC3, C4-2, DU145, multiple myeloma MM.1S cell line, chronic myeloid leukemia K562 cell line, osteosarcoma cell lines U20S and 143B, ovarian cancer cell line ES2 and OVCAR8, pancreatic cancer cell line PANC1 and CAPAN-2, lung cancer cell line H292, colorectal carcinoma cell line HT29 and murine melanoma cell line B16 F10 were obtained from the American Type Culture Collection (ATCC). Pancreatic cancer cell lines PANC28, liver cancer cell line Huh7 and gastric cancer cell line HGC-27 were obtained from the cell bank of Chinese Academy of Sciences. Primary prostatic CAF cells were provided by Changhai Hospital (Shanghai, China), and the passage number of stromal cells used in this work was limited to 3 to 10. PC3, DU145, CAF, Huh7, U20S, 143B, ES2, OVCAR8, PANC1, PANC28, CAPAN-2, HT29, B16 F10 cells were cultured in DMEM medium (Gibco) supplemented with 10% FBS and 1% P/S (100 U/mL penicillin and 100 mg/mL streptomycin). C4-2, MM.1S, K562, H292, HGC-27 cells were cultured in RPMI 1640 (Gibco) supplemented with 10% FBS and 1% P/S. The cell line PC3-PSMA+ which means prostate specific membrane antigen (PSMA) was overexpressed in PC3 cells by lentivirus, and screened with puromycin dihydrochloride (2–4 µg/mL; BBI Life Science, F118BA0026) for stable overexpression of PSMA and luciferase (For the lentiviral plasmid designed for luciferase overexpression, the puromycin resistance gene (PuroR) was co-expressed via P2 A sequence). To obtain other tumor cell lines overexpressing luciferase, we infected luciferase virus and screened using puromycin. CD155 knock out was performed by electroporation with sgRNA and cas9 protein RNP (Ribonucleoprotein) in PC3-PSMA+ cells, CD155 sgRNA (5′-CAGAGAUGGAUACCUCUGGG-3′. Both PC3-PSMA+ and CD155 knock out PC3-PSMA+ cell lines were expanded from single clones obtained by limiting dilution. All the cells maintained in a humidified atmosphere containing 5% carbon dioxide (CO2) incubators (Thermo) at 37 °C and routinely tested to confirm the absence of Mycoplasma contamination using the MycAway Plus-Color One-Step Mycoplasma Detection Kit (Yeasen Bio-technol).

Plasmid construction

We constructed a second-generation CAR. The nucleic acid sequence of the CAR comprised of the CD155 antibody scFv (CN1777622 A: SCFV1) linked in-frame to the hinge and transmembrane regions of the human CD8α chain and intracellular human 4-1BB (CD137) and CD3ζ signaling domains, which was transferred to the empty vector (lentiviral transfer vector pELPS) using XbaӀ and SalӀ double enzymatic digestion. Additionally, we designed a TIGIT ECD (extracellular domain, AA 22–141) based CAR sequence without a hinge region. The sequences of CD155 antibody scFv and TIGIT ECD are listed in Additional file 2: Table S1. All sequences were synthesized by GenScript. Anti-PSMA CAR (Pbbz CAR-T) plasmid was constructed as our previous work [21].

Preparation of CAR-T cells

For lentivirus production, 293 T cells were used to produce lentivirus which carried CAR. 293 T cells were cultured to a degree of 80%−90% and co-transfected with three-plasmid system (packaging vectors psPAX2, envelope plasmid pMD2.G and CAR plasmid) by using polyethyleneimine (Polysciences, PEI MAX 40000). Renewed medium after 6–8 h of post-transfection and virus-containing supernatant was harvested and concentrated. To prepare CAR-T cells, peripheral blood was obtained from healthy volunteers and statements that informed written consent of all participants were obtained. All blood samples were provided by BRL Medicine Inc. Lymphocyte separation medium (TBD Science, LTS10770015) was applied to obtain Peripheral blood mononuclear cells (PBMCs) which were further treated with CD4/CD8 T Cell Isolation Kit (Miltenyi) to get CD4+/CD8+ T cells. After 2 days of activation with T Cell TransAct (Miltenyi), T cells were infected by the lentivirus which mixed evenly with Polybrene at MOI = 10 by spinning at 1800 rpm for 2 h at 37 °C in a prewarmed centrifuge. The expression of CAR was analyzed after two or three days, and the CAR-T cells were used for further experiments.

Human samples

Fresh peripheral blood of metastatic castration resistant prostate cancer (mCRPC) patients was obtained from Changzheng Hospital (Shanghai, China) and statements that informed written consent of mCRPC patients were obtained. All samples were collected and handled according to the ethical and safety procedures approved by the Ethics Committee of the Changzheng Hospital. Human tissue microarrays were purchased from Shanghai Outdo Biotech Co., Itd.

Cytotoxicity assay

The cytotoxic activity of CAR-T cells was assessed using a luciferase-based killing assay. Specifically, target tumor cells expressing firefly luciferase were co-cultured with CAR-T cells or non-infected control T cells at various effector:target ratios in triplicate in 96-well plates (Corning) using 100 μL of cell media. Following a period of co-incubation, 10 μL of fluorescein substrate (Invitrogen) was added per well and luminescence was measured using a microplate reader (BMG LABTECH SPECTROstarNano, 562 nm) and an IVIS instrument (IVIS® Lumina III). Percent lysis was determined by constructing a standard curve through linear regression of luminescence values against the viable number of tumor cells.

Repetitive tumor cell challenge assay

Tumor cells (PC3-PSMA+ cells) and effector cells (PVRbbz CAR-T cells or CTRL-T cells) were initial co-cultured at the ratio of 1:1 (0.5 million CAR+ cells: 0.5 million tumor cells). After co-culture for 48 h, all tumor cells and T cells were collected and counted by flow cytometry. Then, the cells were re-challenged with fresh 0.5 million tumor cells every 48 h. The addition of tumor cells was performed after each count, and the experiment was concluded when a decline in T cell expansion was observed. To monitor the changes in TIGIT and CAR expression, analyses were conducted on the first and last days of the experiment. Furthermore, the expression of CD155 in tumor cells was analyzed at the experimental endpoint.

Flow cytometry

Cells were stained with fluorochrome-conjugated antibodies against the following markers according to the manufacturer’s instructions. PE-conjugated anti-human CD155 (BioLegend, 337,609), PerCP/Cyanine5.5-conjugated anti-human CD15 (BioLegend, 323,019), APC-conjugated anti-human CD19 (BioLegend, 363,005), FITC-conjugated anti-human CD34 (BioLegend, 343,603), FITC-conjugated anti-human CD14 (BioLegend, 301,803), FITC-conjugated anti-human CD3 (BioLegend, 317,306), PerCP/cy5.5-conjugated anti-human CD4 (BioLegend, 300,529), APC/Fire™ 750-conjugated anti-human CD8 (BioLegend, 344,746), PE-conjugated anti-human TNF-α (BioLegend, 502,908), PE-conjugated anti-human IFN-γ (BD Biosciences, 554,552), PE-conjugated anti-human GranzymeB (BD Biosciences, 561,142), PE/Cyanine7-conjugated anti-human TIM-3 (BioLegend, 345,013), PE-conjugated anti-human PD1 (BioLegend, 329,905), PE/Cyanine7-conjugated anti-human PD1 (BioLegend, 329,917), PE-conjugated anti-human LAG-3 (BioLegend, 369,205), PE-conjugated anti-human TIGIT (BioLegend, 372,703), APC-conjugated anti-human CD57 (BioLegend, 359,609), PE-conjugated anti-human CCR7 (BioLegend, 353,204), BV421-conjugated anti-human CD45RA (BioLegend, 304,130), PE-conjugated Mouse IgG1 κ Isotype Ctrl (BioLegend, 400,112), Zombie UV™ Fixable Viability Kit (BioLegend, 423,108), Precision Count Beads™ (BioLegend, 424,902), PE-conjugated anti-human PSMA (BioLegend, 342,503), PE-conjugated anti-human CD69 (BioLegend, 310,905), APC-conjugated anti-human CD25 (BioLegend, 385,605), APC-conjugated anti-mouse TAGE4 (BioLegend, 131,509), APC Rat IgG2a, κ Isotype Ctrl Antibody (BioLegend, 400,511). For intracellular staining, the cells were stimulated with Cell Activation Cocktail (with Brefeldin A) (BioLegend, 423,304) for 6 h, fixed, permeabilized, and then stained with relative antibodies. For scFv based CAR expression detection, cells were stained with biotinylated human CD155/PVR protein (Sino Biological), followed by APC-conjugated Streptavidin (SA). Non-transduced T cells were employed as a control and underwent the same experimental procedure. Flow cytometry was performed on Fortessa-LSR II cytometer (Becton–Dickinson) and analyzed with FlowJo X 10.0.7r2 (Tree Star).

ELISA

IFN-γ and TNF-α were measured by Human IFN-γ (Invitrogen, 88-7316-88), Human TNF-α (Invitrogen, 88-7346-88) ELISA kits following the manufacturer’s instruction, respectively.

Treatment of tumor-bearing mice

Subcutaneous Xenograft model: We used the immunodeficient nude mice for the xenograft studies. Each individual nude mouse (6–8 weeks old) was subcutaneously injected with 1 × 106 human Huh7 cells to establish a liver cancer model. After 14 days of tumor model establishment, further experiments were conducted. The mice were divided into two groups based on an average tumor size of approximately 200 mm3 prior to treatment. Each group (CTRL-T/PVRbbz CAR-T) received the same number of T cells respectively through tail vein injection (5 × 106 cells/mouse, approximately 60% CAR positive) on day 0. Recipient mice were observed daily, tumor size was measured using digital calipers, luminescence was detected by In Vivo Imaging Systems (PerkinElmer) twice weekly, and bodyweight was measured twice a week. All experiments were carried out in accordance with current national and institutional regulations and ethical guidelines, and were performed by an accredited person. Tumor volume was calculated as length × width × width × 0.5 and mice were humanely sacrificed if the tumor volume exceeded 2000 mm3. Mice were euthanized by exposure to carbon dioxide (CO2), followed by cervical dislocation to ensure death.

Osteosarcoma model: To establish an intratibial injection model, a suspension of 143B cells in PBS at a density of 1 × 106 cells/10 μL, and 10 μL of cell solution was slowly injected using a 28.5-G needle with a drilling motion into the tibia. After 15 days, T cells (CTRL-T/PVRbbz CAR-T) were infused intravenously at a dose of 5 × 106. Mice were monitored twice a week using In Vivo Imaging Systems. We also measured tumor volume using calipers to determine whether the tumor reached the ethical endpoint. Tumor volume was calculated as length × width × width × 0.5 and mice were humanely sacrificed if the tumor volume exceeded 2000 mm3.

Hematological tumor model: MM.1S cells (5 × 105 cells per mouse) were suspended in PBS and injected into nude mice (6–8 weeks old) through tail vein. Two weeks after tumor engraftment, CTRL-T or PVRbbz CAR-T cells (1 × 107 cells/mouse) were infused intravenously. Tumor size and bodyweight were monitored every 2–3 days following T cell infusion with In Vivo Imaging Systems.

Immunohistochemistry

Immunohistochemistry (IHC) analysis of human tissue microarrays was performed by Shanghai Outdo Biotech Co., Itd (CD155/PVR antibody was bought from Abcam, NBP1-8813, 1:200 dilution). The scoring of staining intensity was evaluated on a four-point scale; 0 for no expression, + indicating low or weak expression in less than 20% of positive cells, +  + indicating moderate expression in 20% to 50% of positive cells, and +  + + indicating high or strong expression in more than 50% of positive cells.

In some experiments, vital organs were surgically excised from mice and subsequently fixed with neutrally buffered 4% formaldehyde, and applied to hematoxylin and eosin (H&E) staining.

Bioinformatic analyses

Data of Pan cancer expression and overall survival of CD155 were downloaded from GEPIA database (http://gepia.cancer-pku.cn/) [22] and UALCAN database (https://ualcan.path.uab.edu/) [23]. TCGA data for validation of overexpression of CD155 in prostate adenocarcinoma (PRAD, normal: n = 52, tumor: n = 499), liver hepatocellular carcinoma (LIHC, normal: n = 51, tumor: n = 373) and kidney renal papillary cell carcinoma (KIRP, normal: n = 32, tumor: n = 289) were downloaded from XENA database (https://xenabrowser.net/) [24]. AML and Healthy bone marrow data were downloaded from BloodSPOT database, and the dataset Leukemia MILE study (212662_at, normal: n = 73, AML: n = 542, https://servers.binf.ku.dk/bloodspot/?gene=PVR&dataset=all_mile) [25] was used to further analysis. To analyze targets expression in normal tissues single cell sequencing data, we collected single cell RNA sequencing data of seven healthy human tissues, including heart (n = 7, 287,269), spleen (n = 5, 94,257 cells), lung (n = 5, 57,020 cells), kidney (n = 3, 14,296 cells), liver (n = 5, 8444 cells), bone marrow mononuclear cells (n = 8, 30,672 cells) and PBMC (merged pbmc3k, pbmc4k, pbmc6k, pbmc8k, frozen_pbmc_donor_a, frozen_pbmc_donor_b and frozen_pbmc_donor_c data from TENxPBMCData package, 41,042 cells). The heart data was sourced from the Broad Institute’s Single Cell Portal database (http://singlecell.broadinstitute.org/single-cell/study/SCP498/transcriptional-and-cellular-diversity-of-the-human-heart; ID SCP498) [26, 27], spleen and lung tissue data were obtained from the https://www.tissuestabilitycellatlas.org website [28]. The kidney tissue data was sourced from the GEO database (https://www.ncbi.nlm.nih.gov/geo/) dataset GSE131685 [29]. The liver tissue data was obtained from dataset GSE115469. AML single cell sequencing dataset was downloaded from the zenodo link (https://zenodo.org/records/5931689, bone_marrow.h5ad, 339,381 cells from 63 donors) accompanying the publication [30]. Lung adenocarcinoma single cell sequencing data was downloaded from dataset GSE131907 (208,506 cells from 58 samples) [31]. Hepatocellular carcinoma single cell sequencing data was visualized in the https://db.cngb.org/PRHCCdb website [32]. To eliminate batch effects in the single cell RNA sequencing, we used the Find Integration Anchors function and processed the data using the R package Seurat (min.cells = 1). We also applied quality control measures to the unfiltered data and normalized it using the Normalize Data function. The Find Variable Features function, Scale Data function, and RunPCA function were then used for further processing, with appropriate parameters selected based on Elbow Plot results for dimensionality reduction clustering. Clusters were calculated using the FindClusters function. Cell annotation was performed based on differential genes of each group and characteristic genes of cells, for PBMC data, we used Reference Mapping approach to annotate cell labels in a query dataset (https://satijalab.org/seurat/articles/multimodal_reference_mapping.html). And the SCP package and VlnPlot/FeaturePlot function was used for visualization purposes [33].

Statistical analysis

All the experiments were performed in triplicate and independently repeated at least two times, representative results were shown. All statistical analyses were performed using GraphPad Prism V.9 software. All data were mean ± SEM. P values were calculated by two tailed unpaired or paired Student’s t test between two samples, and multiple-comparison analyses were evaluated by ANOVA (one or two way) or Multiple t tests. For survival analysis, P values for Kaplan–Meier curves were determined using the log-rank test. P < 0.05 were deemed as significant.