SW480, HCT116, K562, and Huh7 were obtained from ATCC. The human Jurkat cell line was maintained in our laboratory, which neither produced IFN-γ nor exerted effector killing functions. These cells were cultured in the RPMI 1640 medium (Gibco, Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) (Biological Industries, ISR). Human breast cancer cell lines MDA-MB-231 and HS578T were obtained from ATCC and cultured in the Dulbecco’s modified Eagle medium (Gibco, Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) (Biological Industries, ISR). The HEK-293T cell line was purchased from ATCC and maintained in a complete growth medium (Gibco, Invitrogen), containing with 10% fetal bovine serum (FBS) (Biological Industries, ISR), 2 mM L-glutamine (Gibco, Invitrogen), and 1 mM sodium pyruvate solution (Gibco, Invitrogen). All the aforementioned cells were cultured in a humidified atmosphere containing 5% CO₂ at 37°C.

The target TR1⁴¹⁹ CAR consisted of TR1⁴¹⁹ scFv, CD28, or 4-1BB co-stimulatory domains and a CD3ζ signaling domain (Gross et al., 1989; Irving and Weiss, 1991). They were obtained by overlap PCR amplification; PCR products were cloned into a lentivirus vector pWPXL and verified by enzymatic digestion and sequencing. In order to produce the lentivirus supernatant, 8 × 10⁵ HEK-293T cells were seeded in 6-well plates. After 24 h, 293T cells were transfected with pMD.2G encoding VSV-G envelope, pSPAX2 lentivirus plasmid, and the CAR-pWPXL plasmids by using the Xfect Transfection Reagent (Takara xfect™), according to the manufacturer’s instructions. The culture supernatants of 293T cells containing lentiviral particles were collected and filtered with a 0.45-μM filter at 48 h after transfection.

Human peripheral blood mononuclear cell (PBMC) samples were obtained from healthy donors isolated by using MACSprep™ PBMC Isolation Kit (Miltenyi Biotec, Germany), according to the manufacturer’s instructions. Written informed consents were obtained from all volunteer. In brief, The PBMCs were activated for 48 h in 24-well tissue culture–treated plates (2 × 10⁶/well) with anti-CD3 mAb (Miltenyi, Biotec, 1 μg/ml) and anti-CD28 mAb (Miltenyi, Biotec, 1 μg/ml) in a complete medium containing 90% RPMI 1640 and supplemented with 10% FBS (Gibco), 200 U/ml IL-2, 25 mM HEPPES, 55 μM 2-M, 100 U/ml penicillin, and 100 μg/ml streptomycin. After 2 days, the 2 × 10⁵ activated T cells were infected with 250 μL of CAR-encoding lentiviral supernatant in a tissue culture–treated 24-well plate. After lentiviral infection for 24 h, the lentiviral supernatant was replaced with a fresh complete medium, and cell culture were maintained at 37°C with 5% CO₂. On the 7th day after transfection, the T cells were collected for subsequent experiments.

For cell surface staining, the lentivirus-infected T cells were harvested and washed with phosphate-buffered saline (PBS) followed by incubation with 1ug/ml recombinant protein TRAIL-R1-Fc (R&D Systems, Minneapolis, MN) at room temperature for 20 min. After being washed, the cells were incubation with 1ug/ml APC-conjugated anti-human IgG secondary antibodies and other cocktail antibodies (Biolegend) at room temperature for 15 min in the dark. After being washed with PBS, the stained cells were analyzed by flow cytometry (BD Biosciences). The following monoclonal antibodies were used with the indicated subtypes: APC-labeled human IgG-Fc (clone HP6017), Percy5.5-labeled anti-CD4 (clone RPA-T4), Alex700-labeled anti-CD8 (clone HIT8a), BV421-labeled anti-CCR7 (clone G043H7), PE-labeled anti-CD45RO (clone UCHL1), AF488-labeled anti PD-1 (clone EH12.2H7), BV605-labeled anti-LAG-3 (clone 11C3C65), and BV785-labeled anti-TIM-3 (clone F38-2E2). All sample data analyzed was done on ≥10,000 events using the FlowJo V 10 data analysis software.

The ability of CAR-T cells to kill tumor cells was determined via calcein AM (CAM, Dojindo) release–based cytotoxic cell assay (Jonsson et al., 1996). In brief, target cells were stained with 10 μM CAM for 30 min at 37°C, followed by terminating the reaction with addition of Dulbecco’s phosphate-buffered saline without calcium (DPBS) buffer. The target cells were washed five times with DPBS. Non-transduced T cells were used to normalize the percentage of CAR-positive cells. Then CAR-T cells and target cells were plated in a 96-well microplate at various effector/target (E:T) ratios at 37°C. After incubation for 6h, the 50 μl supernatant was transferred to 96-black plates to measure fluorescence intensity (FI) at 485 nm excitation and 520 nm emission wavelengths. Three parallel holes were set for each group. The percentage of cytotoxicity was calculated according to the following formula: l y s i s % = ( test   release − spontaneous   release ) / ( maximal   release − spontaneous   release ) × 100 % .

The 293T cell lysis by CAR-T cells was assessed by using real-time cell analysis (RTCA, ACEA)–based killing assay. One prior day to seed approximately 1 × 10⁴ 293T cells on the 16 E-plate, the cells were grown to mid-logarithmic growth phase. Non-transduced T cells were used to normalize the percentage of CAR-positive cells, and CAR-T cells were added to the cultures at indicated effector-to-target ratio. A RTCA DP analyzer was used to monitor real-time target cell growth, and the results were analyzed by RTCA software (Bilandzic et al., 2019). The percentage of cytotoxicity was calculated according to the following formula: l y s i s % = [ 1 − ( experiments / empty   culture ) ] × 100 % .

The CAR-T cells and target tumor cells were cocultured at s various E:T ratios in a 96-well plate for 24 h at 37°C. The culture supernatants were harvested and subjected using ELISA to detect IFN-γ and granzyme B, according to the manufacturer’s instructions. Every value represented the mean of triplicate wells.

CAR-T cells and target cells were incubated for 2 h at 37°C, and target cell apoptosis was assayed using an Annexin V/7-AAD Apoptosis Detection Kit (Biolegend), according to the manufacturer’s instructions. Target cell caspase-3 activity was detected used a caspase-3 activity detection assay kit (BD). In brief, the cells were washed twice with cold PBS, followed by fixing and permeabilizing in Cytofix/CytoPerm for 20 min on ice (BD Biosciences). Subsequent staining was performed with 1 × PermWash as the staining and wash buffer and incubated for 15 min at room temperature. Washed cells were analyzed with BD Celesta (BD Biosciences) software by flow cytometry.

Statistical analyses in this study were performed with GraphPad Prism software 8.0.2 version. Data were presented as mean ± SEM. When there were two experiment groups, Student’s t test was used or ANOVA analysis for multiple comparisons. p values < 0.05 were considered statistically significance.

In order to determine the functionally optimal CAR for the TR1⁴¹⁹ scFv structural design, we generated three different TR1⁴¹⁹ CARs with distinct intracellular compartment. Lentiviral vectors were constructed with the scFv region of TR1⁴¹⁹, which were paired with the CD28, the 4-1BB, or the CD28-4-1BB co-stimulatory domain and combined with a CD3ζ activation domain (Supplementary Figure S1). Such vectors were transduced into the PBMCs derived from healthy donors to obtain the human TR1⁴¹⁹-28ζ, TR1⁴¹⁹-BBζ, and TR1⁴¹⁹-28BBζ CAR-T cells, respectively. A flow cytometric analysis confirmed the similar surface expression levels of corresponding CARs on each type of T cells (Figure 1A). We checked for the phenotypes of these T cells and found that naive T cells constituted approximately 60% of each CAR-T cell populations (Figure 1B). There were no apparent variations in the different subtype proportions shown among the three types of CAR-T cells (Figure 1B). Moreover, we analyzed the exhaustion markers of these T cells, and no significant differences of PD-1, LAG-3 and TIM-3 expression were observed among the TR1⁴¹⁹-28ζ, TR1⁴¹⁹-BBζ, and TR1⁴¹⁹-28BBζ CAR-T cells (Figure 1C, Supplementary Figure S2). These data demonstrated that there were no marked differences in the T cell phenotypic features between introducing the second or the third generation of TR1⁴¹⁹-CARs.

Next, we examined the cellular activities of TR1⁴¹⁹-28ζ, TR1⁴¹⁹-BBζ, and TR1⁴¹⁹-28BBζ CAR-T cells upon antigen stimulation. After 24 h of stimulation with different concentrations of plate-bound recombinant human TRAIL-R1 protein, the expression of CD137 was shown to be significantly higher in TR1⁴¹⁹-28BBζ CAR-T cells at different antigen concentrations (Figure 1D). In addition, the T-cell proliferation was studied using CFSE staining, and the cell proliferation was analyzed by quantifying the fluorescent intensities of CFSE (cells with low CFSE peaks are proliferating cells) (Hadjadj et al., 2020). We found that the number of all three types of CAR-T cells was enlarged after cocultured with TRAIL-R1–positive SW480 cancer cells for 7 days (Figure 1E). Among them, TR1⁴¹⁹-28BBζ CAR-T cells showed superior expansion capability compared with TR1⁴¹⁹-28ζ and TR1⁴¹⁹-BBζ CAR-T cells, and no significant difference was detected in cell proliferation between the 2 second-generation CAR-T cells (Figure 1E). A comparably higher level of PD-1 expression was found in TR1⁴¹⁹28BBζ CAR-T cells than in TR1⁴¹⁹-28ζ or TR1⁴¹⁹-BBζ CAR-T cells after co-incubation with SW480, which was in line with the findings shown before that PD-1 was more frequently expressed by the TR1⁴¹⁹-28BBζ CAR-transduced T cells (Figures 1F,G). Collectively, these results showed that the co-stimulatory domain of the TR1⁴¹⁹-28BBζ CAR was associated with significantly higher levels of CAR-T cell activation and proliferation upon antigen stimulation.

In order to study the cytotoxic efficacy of TR1⁴¹⁹-28BBζ CAR-T cells against target tumor cells in vitro, we screened for a panel of TRAIL-R1–positive cancer cell lines (Supplementary Figure S3). Results from the calcein AM release–based cytotoxic cell assay showed that TR1⁴¹⁹-28BBζ CAR-T cells exhibited profound potency in killing TRAIL-R1–expressing SW480, HCT116, or MDA-MB-231 cells than the mock T cells at all E:T ratios (Figure 2A). Such cytolytic effects were accompanied by significantly increased IFN-γ and granzyme B secretions (Figures 2B,C). In comparison, the TRAIL-R1–negative 293T cells showed no susceptibility toward TR1⁴¹⁹-28BBζ CAR-T cells in the coculture system, as no significant differences in cell lysis or effector cytokine secretions were observed comparing with the mock T group. Also, we found that TR1⁴¹⁹-28ζ CAR-T cells could effectively kill SW480 and HCT116 tumor cells, with the efficiency similar as TR1⁴¹⁹-28BBζ CAR-T cells, while the cytolytic efficacy of TR1⁴¹⁹-BBζ was comparatively lower than that of other CAR-T cells against HCT116 cells (Supplementary Figure S4). These data demonstrated that the engineered TR1⁴¹⁹-28BBζ CAR-T cells exhibited high functional efficacy against multiple tumor cells expressing the surface receptor TRAIL-R1.

To access the TRAIL-dependent cytotoxic effect with the TR1⁴¹⁹-28BBζ CAR-T cells, we transduced the third-generation TR1⁴¹⁹ 28BBζ CAR into Jurkat cells, which were unable to generate regular T cell cytolytic functions. In addition, we constructed a TR1⁴¹⁹∆ζ lentiviral vector, carrying the truncated TR1⁴¹⁹ CAR with no co-stimulatory domain or CD3ζ domain to completely blunt T-cell activation (Hayes et al., 2002; Yudushkin and Vale, 2010) (Supplementary Figure S1). The flow cytometry analysis confirmed that the expression of TR1⁴¹⁹ scFv could be detected on the surface of both TR1⁴¹⁹-28BBζ CAR- and TR1⁴¹⁹∆ζ CAR–expressing Jurkat cells (Figure 3A). Target cell death was monitored using the RTCA based killing assay, and we found that TR1⁴¹⁹-28BBζ and TR1⁴¹⁹∆ζ-Jurkat cells could both efficiently induce SW480 tumor cell apoptosis in less than 24 h of coculturing. Furthermore, we utilized the soluble recombinant human TRAIL-R1-Fc proteins to block the scFv region expressed on both TR1⁴¹⁹-28BBζ and TR1⁴¹⁹∆ζ CAR-T cells, in order to confirm whether the extracellular domain of TR1⁴¹⁹ CAR could induce tumor cell death. We found that co-treatment of such soluble TR1⁴¹⁹ scFv blockers could completely abrogate the cytolytic effect of both types of CAR-Jurkat cells on the SW480 target cells (Figure 3B).

To determine the activation of the apoptosis pathway downstream of the target antigen, we performed the flow cytometric analysis for the expression of Annexin Ⅴ/7-AAD and caspase-3/cleaved caspase-3 in SW480 cells. It was shown that the percentages of apoptotic SW480 cells were similar as 45.7 and 40.5% for the TR1⁴¹⁹-28BBζ-Jurkat cells and TR1⁴¹⁹∆ζ-Jurkat cells cocultures, respectively (Figure 3C). These CAR-Jurkat cells induced levels of apoptosis were significantly higher than the control mock-Jurkat cells as well as the cocultures mixed with soluble target antigen recombinant proteins (Figure 3C). Similar results were found for Caspase-3 expression in SW480 cells (Figure 3D). Collectively, these results suggested that the scFv region from the TR1⁴¹⁹ CAR played an independent role in the cytolytic effect of the CAT-T cells as it could mediate apoptosis in the TRAIL-R1–positive tumor cells.

To further confirm TR1⁴¹⁹-28BBζ CAR-T cells exhibited dual antitumor effects through death receptor TRAIL-R1–mediated apoptosis as well as T cell induced cytolysis, we established two target lines with 293T cells. 293T cells were exogenously introduced with the wild type receptor or a truncated TRAIL-R1 missing the death domain that was responsible for intracellular activation of the apoptosis pathway (Schneider et al., 1997) (Supplementary Figure S5). In parallel, we generated TR1⁴¹⁹∆ζ-T cells by lentiviral transduction, to be used as a negative control for the TR1⁴¹⁹-28BBζ CAR-T cells (Figures 4A,B). Then we performed cytotoxicity assays using TR1⁴¹⁹-28BBζ CAR-T cells and TR1⁴¹⁹∆ζ CAR-T cells as effector cells, cocultured with the 293T target system. For the 293T cells expressing the full-length TRAIL-R1, the most efficient level of cell killing was achieved with the TR1⁴¹⁹-28BBζ CAR-T cells. Blocking the T-cell activation signal could reduce the overall level of cell lysis only at higher E:T ratios. In 293T cells with modulations on the integrity of the death receptor pathway, TR1⁴¹⁹-28BBζ CAR-T cells could induce cell lysis in high efficiency, but such cytotoxic effect was completely abolished when T-cell activation was interrupted in the TR1⁴¹⁹∆ζ CAR-T cells (Figure 4C). Because truncating the DD intracellular region of the antigen would not affect the CAR recognition, the production of IFN-γ and granzyme B were remained at a relatively high level in both 293T (full-length) and 293T (truncation) cells, whereas they were dropped to barely detectable levels when cocultured with TR1⁴¹⁹∆ζ CAR-T cells (Figures 4D,E). These results suggested that TR1⁴¹⁹ CAR-T cells exhibited cell killing efficacy not only by inducing T cell–mediated cytolysis of tumor cells but also by promoting tumor cell death through activation of the TRAIL-R1–dependent apoptotic pathway.