CD70 is a type II transmembrane protein that belongs to the tumor necrosis factor (TNF) family. As an immunoregulatory molecule, it is typically expressed on activated T cells and B cells and functions to promote the proliferation, differentiation, and survival of lymphocytes (28). CD70 is also highly expressed in renal cancer, glioblastoma, and ovarian cancer tissues, as demonstrated by analyses from the Cancer Genome Atlas (TCGA) database ( Supplementary Figure S1A ) and immunohistochemical (IHC) staining results ( Supplementary Figure S1B ), along with findings from previous studies (21, 22, 29). This elevated expression positions CD70 as a promising target for anti-cancer immunotherapy. Therefore, we chose CD70 as the target for constructing CAR-T cells designed to treat solid tumors. We identified tumor cell lines with high CD70 expression, such as U251, ACHN and 786-0, as positive models, while CASKI and BXPC3 served as negative controls for further evaluation ( Supplementary Figure S1C ).

To construct anti-CD70 CAR-T cells, single-chain variable fragments (scFvs) targeting the extracellular domain of CD70 were generated by screening a human phage display antibody library in our laboratory, following previously described methods (data not shown) (30). The screened anti-CD70 scFvs were inserted into a CAR structure to create various anti-CD70 CARs (CD70 CARs) ( Figure 1A ).

To evaluate the affinity of CD70 CARs for CD70 on tumor cells, Jurkat reporter cells engineered to express CD70 CARs were co-cultured with CD70-positive 786-0 cells. In the Jurkat reporter cells, a luciferase reporter gene was placed under the control of a nuclear factor of activated T cells (NFAT)-responsive promoter. Upon antigen binding and CAR activation, NFAT would be dephosphorylated, and the NFAT-responsive promoter would be activated to promote the expression of the luciferase reporter gene. The luminescence intensity of the luciferase signal was measured, reflecting the strength of NFAT-mediated signaling and, indirectly, the affinity and functional activity of the CD70 CAR. Jurkat cells expressing the top 25 CD70 CARs based on luminescence intensity were selected for further verification ( Figure 1B ).

To assess the specificity of the CD70 CARs, the CAR-expressing Jurkat reporter cells were co-cultured with both 786-0 cells and CD70 knockout (KO) 786-0 cells. The luminescence intensity was measured, and the ratios of luminescence intensity between the two groups were calculated. The top five CD70 CARs with the highest ratios were then selected for further functional studies ( Figure 1C ).

The top five scFvs selected from Figure 1C were then incorporated into CD70 CAR-T cells, and the short-term tumor-killing ability of these CD70 CAR-T cells was assessed by co-culturing them with 786-0 cells for 6 hours. The CAR sequence P8F8, screened by Pfizer Inc. (Patent number: US 2023/0041456 A1), was used as a positive control (29). The results demonstrated that the killing efficiency of all five constructed CAR-T cells were comparable to or even superior to that of P8F8 CAR-T cells ( Figure 1D ).

To evaluate the potential cross-reactivity of the selected scFvs with other membrane proteins, a membrane proteome array (MPA) assay was conducted. The results revealed that, apart from the Fc fragments of IgG receptors (FCGR1A, FCGR2B, and FCGR3B), which served as positive controls, the antibodies constructed with P025, A034, G060, and E066 scFvs exclusively bound to CD70. In contrast, the antibody constructed with the C020 scFv interacted with multiple membrane proteins, including CD70, ADGRF1, and ATP11C ( Figure 1E ). Consequently, the scFvs with better specificity, P025, A034, G060, and E066, were selected for further in vivo functionality tests. ACHN and 786-0 xenograft mouse models were established through subcutaneous injection, and the results of CAR-T infusion showed that E066 CAR-T cells outperformed the other CAR-T cells in tumor elimination in both models ( Figures 1F, G ).

Since CD70 is also expressed on activated T cells, CD70 CAR-T cells may recognize and target CD70 on these activated T cells, potentially leading to fratricide (31, 32). To assess fratricide among the generated CD70 CAR-T cells, we conducted a co-culture assay using CFSE-labeled non-CD70-targeting CAR-T cells as target cells and unlabeled CD70 CAR-T cells as effector cells, and the effector-target ratio was set as 1:1. The expression of CD70 on CFSE-labeled cells was measured by flow cytometry before and after co-culture. In this assay, a reduction in CD70-positive cells would indicate fratricide. Cells co-cultured with E066 CAR-T cells exhibited a 5% reduction in CD70 positivity, which was notably lower than the reduction observed in cells co-cultured with P8F8 CAR-T cells ( Figure 1H ). Additionally, the apoptotic rate of CFSE-labeled cells after co-culture with E066 CAR-T cells was approximately 7%, slightly lower than the rate observed in cells co-cultured with P8F8 CAR-T cells ( Figure 1I ). These results indicate that E066 CAR-T cells exhibit minimal fratricide compared to P8F8 CAR-T cells. Consequently, the E066 scFv was selected, leading to the construction of CD70 CAR-T cells with high specificity, strong anti-tumor efficacy, and fratricide-resistance.

Given the limitations of autologous CAR-T therapy, such as high costs and instability, allogeneic CAR-T cells are being developed. The main challenges associated with allogeneic CAR-T therapy include graft-versus-host disease (GVHD) and host-versus-graft (HVG) reactions, which arise from the TCR and HLA molecules present on the surface of CAR-T cells (7). To inhibit TCR expression, we knocked out the TCR-encoding gene TRAC using CRISPR/Cas9 technology, while the levels of HLA molecules were reduced by silencing B2M and HLA-DRA ( Figure 2A ). For each gene, 10 single-guide RNAs (sgRNAs) were designed, and the knockout efficiency was assessed using flow cytometry analysis ( Supplementary Figure S2A ). The sgRNA with the highest knockout efficiency for each gene was selected.

The T7 endonuclease I (T7E1) assay was also conducted to assess the editing efficiency achieved by the CRISPR/Cas9 system. T7E1 specifically cleaves mismatched DNA strands, enabling the identification of bands corresponding to edited and unedited alleles. The editing efficiency was quantified by comparing the intensity of the cleaved and uncut products. The results indicated successful knockout of TRAC, B2M, and HLA-DRA, with efficiencies of 70.4%, 74.2%, and 66.3%, respectively ( Supplementary Figure S2B ).

To evaluate the off-target risks associated with the sgRNAs used and ensure the safety of the knockout system, oligodeoxynucleotide (ODN) tags were integrated into CAR-T cells, followed by high-throughput sequencing of the PCR-amplified DNA fragments. The off-target rates for these sgRNAs were found to be 8.97%, 2.70%, and 0% for TRAC, B2M, and HLA-DRA, respectively ( Supplementary Figure S2C ). Subsequently, we performed a triple knockout (tKO) of TRAC, B2M, and HLA-DRA in CD70 CAR-T cells to generate CD70 UCAR-T cells, achieving knockout efficiencies of over 80% for each gene ( Figure 2B ).

To examine the GVHD induced by UCAR-T cells, we co-cultured UCAR-T cells and unmodified CAR-T cells (referred to as mock CAR-T cells) with sublethally irradiated human peripheral blood mononuclear cells (PBMCs) obtained from a different donor. Compared to mock CAR-T cells, UCAR-T cells exhibited less activation upon PBMC stimulation, as indicated by changes in CD69 expression ( Supplementary Figure S3A and Figure 2C ). The cytotoxicity of the allogeneic CAR-T cells against PBMCs was assessed by measuring the apoptotic rate of PBMCs. Mock CAR-T cells demonstrated a significantly higher killing rate against PBMCs, while UCAR-T cells induced minimal killing ( Figure 2D ). These results indicate that the GVHD potential of UCAR-T cells has been reduced due to the knockout of TCR in CAR-T cells.

Mock CAR-T cells, TRAC single knockout CAR-T cells (sKO), TRAC and B2M double knockout CAR-T cells (dKO), and TRAC, B2M, and HLA-DRA triple knockout CAR-T cells (tKO) were co-cultured with host CD4⁺ and CD8⁺ T cells to assess HVG reactions. Activation of host T cells was evaluated by measuring CD69 levels. Both CD4⁺ and CD8⁺ T cells co-cultured with UCAR-T cells exhibited less activation compared to the other groups ( Supplementary Figures S3B, E ). Additionally, host T cells demonstrated lower cytotoxicity against UCAR-T cells, indicating a significant reduction in the risk of triggering HVG responses ( Figures 2E, F ).

In addition to T cells, host NK cells and macrophages can also target allogeneic CAR-T cells. We observed that when UCAR-T cells were co-cultured with NK cells, their apoptosis increased due to the knockout of HLA-I ( Figure 2G ), consistent with previous studies (16). Furthermore, while macrophages induced less damage to UCAR-T cells compared to mock CAR-T cells due to the knockouts of HLA molecules, their cytotoxicity was still significantly higher than that observed in autologous CAR-T cells derived from the same donor as the macrophages ( Figure 2H ).

Since CD47 has been reported to facilitate the immune escape from macrophages and NK cells and protect CAR-T cells (33–35), we overexpressed CD47 in UCAR-T cells and constructed CD47 UCAR-T cells, aiming to reduce HVG reactions of UCAR-T cells ( Figure 3A ). The expressions of the CD70 CAR and CD47 were validated through flow cytometry analysis ( Figure 3B ). Subsequently, the CAR-T cells were co-cultured with NK cells, and NK cell activation, indicated by the upregulation of CD107a, as well as CAR-T cell apoptosis, were measured. The results showed that compared to TRAC KO CAR-T cells, the knockout of HLA molecules in UCAR-T cells led to increased NK cell activation, while the expression of CD47 mitigated this increase ( Figures 3C, D ). Consistently, upon co-culture with NK cells, the apoptotic rate of CD47 UCAR-T cells was lower than that of UCAR-T cells, suggesting that CD47, acting as an inhibitory ligand, reduced NK cell reactivity ( Figure 3E ).

Similarly, to assess whether CD47 UCAR-T cells can withstand macrophage killing, macrophages and UCAR-T cells were labeled with different fluorescent dyes and co-cultured at an effector-to-target (E: T) ratio of 1:1. Following this, cells were analyzed using flow cytometry. A lower percentage of dual-fluorescent cells in the CD47 UCAR-T group indicated a reduced phagocytic rate of macrophages ( Figures 3F, G ). Additionally, a significantly decreased apoptotic rate of CD47 UCAR-T cells was observed, highlighting the crucial role of CD47 in protecting UCAR-T cells from macrophage attacks ( Figure 3H ).

Another common concern regarding the application of UCAR-T cells is their inadequate in vivo survival and persistence resulting from gene knockouts (17). During in vitro expansion, we observed that the survival and proliferation ability of UCAR-T cells were significantly restricted compared to mock CAR-T cells ( Figure 4A ), as was their tumor-killing capability at a lower E:T ratio ( Figure 4B ). Additionally, replacing IL-2 with IL-7 and IL-15 in the culture media showed beneficial effects for the long-term survival and anti-tumor function of UCAR-T cells.

Considering that the extracellular domain of IL-15 closely resembles that of IL-2, we propose that enhancing IL-7 signaling in UCAR-T cells may facilitate the expansion and function of UCAR-T cells. To activate IL-7 signaling independently of IL-7, we introduced a mutant form of the IL-7 receptor (IL-7R) transmembrane domain, which has been reported to drive constitutive signaling via JAK1 (36, 37), along with the intracellular domain of IL-7R into UCAR-T cells. Together with the extracellular domain of CD47, we constructed and expressed a self-activated and protective (SAP) module in UCAR-T cells to generate SAP UCAR-T cells ( Figure 4C ).

The expressions of the CD70 CAR and CD47 were validated using flow cytometry analysis ( Figure 4D ). The activation of IL-7 signaling was confirmed by the upregulation of phosphorylated STAT5 (pSTAT5), a downstream marker of IL-7R signaling ( Figure 4E ). We then assessed the expansion capabilities of various modified CAR-T cells. Without cytokine stimulation, mock CAR-T, UCAR-T, and CD47 UCAR-T cells failed to expand, while SAP UCAR-T cells retained substantial expansion capability ( Figure 4F ).

The upregulation of pro-survival and proliferation molecules was also observed in SAP UCAR-T cells. ZAP70 is a key molecule involved in T cell survival and proliferation (38, 39), and flow cytometry analysis showed that the level of pZAP70 in UCAR-T cells was reduced due to TCR knockout, subsequently inhibiting the function of UCAR-T cells ( Figure 4G ). Following the expression of the SAP module, the level of pZAP70 was upregulated in SAP UCAR-T cells. Additionally, Western blotting results showed significantly elevated levels of activated STAT5 and BCL2 in SAP UCAR-T cells under cytokine-free conditions ( Figures 4H–J ).

Next, NKG mice were injected with human PBMCs to mimic the human immune system, after which CAR-T cells were infused ( Figure 4K ). Compared to mock CAR-T, UCAR-T, and CD47 UCAR-T cells, the number of SAP UCAR-T cells in the blood was significantly higher, indicating reduced HVG reactions and enhanced persistence ( Figure 4L ). Additionally, CAR-T cells were infused into tumor-free immunodeficient mice to examine their in vivo survival and proliferation in a stimulation-free environment. It was observed that SAP UCAR-T cells expanded for approximately 14 days and remained detectable until Day 42. In contrast, CAR-T cells lacking the SAP module could not proliferate, and their numbers rapidly declined after infusion ( Figure 4M ). In summary, SAP UCAR-T cells with low immunogenicity, resistance to immune rejection, and superior survival and proliferation capabilities were successfully constructed both in vitro and in vivo.

To investigate the anti-tumor function of SAP UCAR-T cells in vitro, the cells were repeatedly challenged with tumor cells for four rounds. Flow cytometry analysis revealed that the expression of the SAP module significantly increased the memory phenotypes in UCAR-T cells ( Figure 5A ). SAP UCAR-T cells also exhibited the lowest levels of exhaustion among the four groups, as indicated by LAG-3 expression ( Figure 5B ). Furthermore, SAP UCAR-T cells demonstrated enhanced proliferation during this process ( Figure 5C ). In terms of anti-tumor efficacy, during the sustained tumor-killing period, SAP UCAR-T cells consistently displayed strong tumor-killing capabilities, comparable to those of mock CAR-T cells ( Figure 5D ). The SAP module also inhibited the release of the immunosuppressive factor IL-10, and promoted the release of tumor necrosis factor-α (TNF-α) and interferon-γ (IFN-γ) in UCAR-T cells to target tumor cells, while maintaining comparable levels of granzyme B release to those observed in the mock CAR-T group ( Figure 5E ). In addition, elevated IL-6 levels are considered as the hallmark of cytokine-release syndrome (CRS) (40), a common side effect of CAR-T therapy, and the reduced IL-6 releasing observed in SAP UCAR-T cells suggested a lower risk of safety issues ( Figure 5E ).

To evaluate the specificity of SAP UCAR-T cells, mock T cells, mock CAR-T cells, and SAP UCAR-T cells were co-cultured with CD70-positive 786-0 cells, as well as CD70-negative CASKI and BXPC3 cells. Cytotoxicity was assessed using a real-time cytotoxicity assay. The results demonstrated that SAP UCAR-T cells effectively killed only CD70-positive 786-0 cells, exhibiting low off-target cytotoxicity ( Figure 5F ). Collectively, SAP UCAR-T cells showed improved expansion, persistence, and anti-tumor activity in vitro, and further in vivo validation is required.

To validate the anti-tumor activity of SAP UCAR-T cells in vivo, a U251 tumor xenograft mouse model was employed. When the tumor volume reached approximately 100 mm³, mice were injected with PBS or 2×10⁶ mock T cells, mock CAR-T cells, UCAR-T cells, or SAP UCAR-T cells ( Figure 6A ). While mock CAR-T and UCAR-T cells significantly inhibited tumor growth, only the SAP UCAR-T group achieved 100% tumor clearance ( Figures 6B, C ). Additionally, the treatment with CAR-T cells prolonged the survival of mice ( Figure 6D ). IHC staining for CD45 in tumor tissues revealed that SAP UCAR-T cells exhibited a stronger infiltration ability into the tumor environment ( Figures 6E, F ).

Mice blood was collected on Day 4, 8, 12, 16 and 28, and real-time quantitative PCR results demonstrated a significantly higher number of CD70 CAR copies in the SAP UCAR-T group ( Figure 6G ). Additionally, cytokine levels in the blood were assessed, revealing that both IFN-γ and TNF-α levels were elevated in the SAP UCAR-T group compared to the mock CAR-T and UCAR-T groups. Conversely, the level of the inhibitory cytokine IL-6 was lower in the SAP UCAR-T group ( Figure 6H ).

A slight weight loss occurred in the mock CAR-T group, potentially due to GVHD and HVG reactions, while no significant changes in mouse weight were observed in other groups ( Figures 6I ). Throughout the CAR-T treatment period, various clinical parameters—such as skin condition, fur quality, body weight, posture, and survival—were statistically analyzed according to the NIH (National Institutes of Health) clinical GVHD scoring criteria (41). The scoring results indicated that neither SAP UCAR-T nor UCAR-T cells produced any noticeable GVHD reactions ( Figure 6J ). Additionally, to assess the toxicity of CAR-T cells on other mouse tissues, hematoxylin and eosin (H&E) staining was performed, revealing no damage to the major tissues in the mice treated with CAR-T cells ( Figure 6K ).

To further assess the safety of SAP UCAR-T cells in vivo, up to 6×10⁶ cells were injected into 786-0 tumor-bearing mice. The successful elimination of tumors confirmed the efficacy of SAP UCAR-T cells in this mouse model ( Supplementary Figure S4A ). Monitoring of body weight ( Supplementary Figure S4B ) and H&E staining of major tissues ( Supplementary Figure S4C ) indicated that the high-dose injection of SAP UCAR-T cells did not raise any safety concerns. Additionally, multiple doses (up to five doses) of SAP UCAR-T cells were administered to the same tumor-bearing mice ( Supplementary Figure S4D ). The results demonstrated effective tumor clearance ( Supplementary Figure S4E ), with no body weight loss or tissue damage observed, even after five doses of injection ( Supplementary Figures S4F, G ). Collectively, these findings suggest that SAP UCAR-T cells possess enhanced tumor-killing ability in vivo without inducing GVHD or causing tissue damage in mice.

To investigate how the SAP module promoted survival, proliferation, and anti-tumor function of UCAR-T cells, RNA sequencing was conducted on mock CAR-T, UCAR-T, and SAP UCAR-T cells. Both volcano plot and cluster analysis revealed significant differences in gene expression between UCAR-T cells and SAP UCAR-T cells, with 92 genes upregulated and 63 genes downregulated in SAP UCAR-T cells ( Figures 7A, B ). KEGG and GO enrichment analyses indicated alterations in lymphocyte reactivity and chemokine signaling pathways attributable to the SAP module ( Figures 7C, D ). Gene set enrichment analysis (GSEA) demonstrated that C-C chemokine receptor activity, JAK-STAT signaling pathways, Th17 cell differentiation and cytokine-cytokine interaction-related genes were significantly upregulated in SAP UCAR-T cells ( Figures 7E–H ).

Notably, X-C motif chemokine ligand 1 and 2 (XCL1 and XCL2), which play a role in leukocyte activation (42, 43), were significantly elevated in SAP UCAR-T cells. In contrast, regulator of G protein signaling 1 (RGS1) and MAX dimerization protein 1 (MXD1), considered as negative regulators of T cell survival, expansion and function (44, 45), were downregulated in these cells ( Figure 7I ).

A complete scale-up workflow was established to evaluate the potential of engineered SAP UCAR-T cells for clinical application ( Figure 8A ), and scale-up production of SAP UCAR-T cells was accomplished by OBiO Technology (Shanghai). In detail, T cells were isolated from the PBMCs of healthy donors and activated using CD3/CD28 magnetic beads. One day post-activation, the activated T cells were transduced with lentivirus, followed by electroporation for gene knockout using the CRISPR/Cas9 system. Residual CD3⁺ T cells were then removed using CD3 positive selection magnetic beads, and cell expansion was conducted in a wave bioreactor. Eleven days after activation, SAP UCAR-T cells were harvested, allocated, and processed for cryopreservation. After undergoing quality assurance (QA) and quality control (QC) testing, the cryopreserved cells were prepared for clinical trials.

We found that the CAR positive rate, cell viability, and proliferation were not affected by large-scale production ( Figures 8B–D ). Additionally, flow cytometry analysis confirmed that knockout efficiency was maintained during the scale-up process ( Figure 8E ). The proportion of memory phenotype SAP UCAR-T cells and the release of IFN-γ were also assessed, showing no significant changes between small-scale and large-scale production ( Figures 8F, G ). In summary, our pre-clinical studies demonstrate that SAP UCAR-T cells exhibit low immunogenicity, satisfactory persistence, and robust anti-solid tumor activity, along with reduced safety concerns both in vitro and in vivo. The stability observed during scale-up production further supports their potential for clinical application.

ACHN (renal cell carcinoma, CRL-1611), and CASKI (cervical cancer, CRM-CRL-1550) cells were purchased from ATCC and cultured in DMEM medium supplemented with 10% fetal bovine serum (FBS). U251 (glioblastoma, CL-0237) cells were obtained from Puno Sai and cultured in DMEM containing 10% FBS. BXPC3 (pancreatic cancer, CRL-1687) cells and 786-0 (renal cell carcinoma, CRL-1932) were sourced from ATCC and cultured in RPMI 1640 medium supplemented with 10% FBS.

The expression of CD70 CAR was analyzed using flow cytometry with FITC-Labeled Human CD27 Ligand/CD70 Protein and His Tag (ACRO, CDL-HF249). PE mouse anti-human CD70 antibody (BD#561935) was employed to detect the expression of CD70 on various cell lines. The phenotypes of CAR-T cells were analyzed using antibodies from BioLegend, including APC anti-human CD3 Antibody (317317), PE anti-human HLA-A,B,C Antibody (311405), FITC anti-human HLA-DR Antibody (327005), APC anti-human CD47 Antibody (323123), PE anti-human CD4 Antibody (300507), APC anti-human CD8 Antibody (344721), APC anti-human CD69 Antibody (310909), APC anti-human CD45 Antibody (304011), APC anti-human CD45RO Antibody (304210), PE anti-human CD62L Antibody (304805), PE anti-human CD223 (LAG-3) Antibody (369205). The CytoFLEX flow cytometer (Beckman Coulter, Inc., California, USA) was then used to acquire flow cytometry data, and the data was analyzed using CytoExpert software.

The second-generation anti-CD70 CAR was designed to include a DNA fragment encoding an in-frame component from the 5’ end to the 3’ end: the CD8α signal peptide, anti-CD70 scFv, CD8 hinge region, CD28 transmembrane domain, and the intracellular co-stimulatory domains of CD28 and CD3zeta. This DNA fragment was codon-optimized using the GenSmart online tool, and synthesized by Anhui General Biotechnology Co., Ltd. (Anhui, China). The overall architecture of the plasmid originated from plasmid pCDH-CMV (Addgene, #72265), which was purchased from System Biosciences (USA). The SAP module was constructed by integrating the extracellular domain of CD47, the transmembrane domain of C7R (a mutant form of IL-7R), and the intracellular domain of IL-7Rα. The transmembrane domain of C7R used in this module is a mutant form (p.Thr244-Ile245insCysProThr) that can be activated independent of the presence of IL-7 according to previous studies (36). The SAP module was linked with the CAR structure by a T2A sequence.

The NFAT-Luc reporter Jurkat cell line was purchased from Jiman Biotechnology Co., Ltd. (Shanghai, China). Various CARs were transduced into the Jurkat cells using lentiviral vectors. Following transduction, the CAR-expressing Jurkat cells were co-cultured with 786-0 cells. The luminescence intensity of the Jurkat cells was measured using a plate reader, serving as an indicator of their activation level in response to CAR activation.

CAR lentivirus was produced by transient transfection of HEK293T cells (CRL-11268, ATCC) using lentiviral plasmids. When the cells reached 80% confluence in T75 flasks, they were co-transfected with the lentiviral plasmid and packaging plasmids pMDLg/pRRE (Addgene, #12251), pRSV-Rev (Addgene, #12253), and pMD2.G (Addgene, #12259) using polyethylenimine (PolyScience, #24765-1). The medium was changed 48 hours post-transfection. Viral supernatants were harvested 48 and 72 hours after transfection, concentrated by ultracentrifugation at 22,000 rpm for 2 hours at 4°C, and stored at –80°C until use.

PBMCs were obtained from HeYou Biotech and MiaoShun Biotech and activated using anti-CD3/CD28 Dynabeads (Gibco) at a 1:1 bead-to-cell ratio in CTS™ OpTmizer™ T Cell Expansion SFM (Gibco), supplemented with 5% human platelet lysate, 5 ng/mL IL-7, and 10 ng/mL IL-15 (Miltenyi Biotec Technology & Trading Co., Ltd.). Twenty-four hours later, the activated T cells were transduced with the CAR lentivirus to generate CAR-T cells.

The membrane proteome array screen was conducted by Integral Molecular, Inc. (USA). In brief, plasmids containing cDNA clones of approximately 6,000 membrane proteins, representing over 94% of the human membrane proteome, were transfected into HEK-293T cells in 384-well plates, with each well containing 18,000 cells. This process generated membrane proteome array matrix plates. Test ligands were subsequently added to these matrix plates and incubated with the membrane protein-expressing HEK-293T cells. After incubation, the cells were washed with PBS and analyzed by flow cytometry using a fluorescence-labeled anti-CD70 antibody to identify the test ligand targets.

Two days post-CAR transduction, CRISPR/Cas9-mediated knockout of TCR and HLA molecules was performed. The gRNAs were mixed with Cas9 protein at a 5:4 ratio and incubated for 10 minutes. Following this, 1×10⁶ cells were electroporated using 20 μL buffer on a 4D-Nucleofector X-unit, according to the manufacturer’s instructions (Lonza). T cells treated with non-binding gRNA served as controls. After electroporation, the T cells were transferred to fresh medium and expanded at a density of 1×10⁶ cells/mL for 10 days.

Genomic DNA was extracted from both unmodified and modified CAR-T cells, followed by PCR amplification targeting specific sites using the primers listed in Supplementary Table S1 . The resulting amplicons were denatured and re-annealed to form heteroduplexes. T7 endonuclease I (T7E1; Nanjing Vazyme Biotech Co., Ltd.) was then used to digest these heteroduplexes, cleaving the mismatched DNA strands. The digested products were analyzed via gel electrophoresis to separate the resulting fragments.

Oligodeoxynucleotide (ODN) tags were introduced during the CRISPR/Cas9-mediated knockout of CAR-T cells, and DNA was extracted from the cells one day later. PCR amplification was conducted using forward and reverse ODN primers. The resulting PCR products were subjected to high-throughput sequencing, and the amplified sequences were compared with the corresponding target sequences.

In the short-term killing assay, CAR-T cells and luciferase-expressing 786-0 (786-0-luc) cells were co-cultured at E:T ratios of 1:1, 3:1, and 9:1 for 6 hours. After incubation, the supernatants were removed, and tumor cells were treated with luciferase substrate for 10 minutes. The luminescence intensity was then detected and measured using a plate reader. The cell killing rate was calculated as [(background luminescence value - sample luminescence value)/background luminescence value] × 100%.

In the long-term killing assay, CAR-T cells were stimulated with 786-0-luc cells at an E:T ratio of 3:1. The CAR-T cells were then rechallenged with 786-0-luc cells for a total of four rounds. After each round, the phenotype and cytotoxicity of the CAR-T cells were assessed.

In the real-time cytotoxicity assay, a 96-well 20idf electrode plate was initially coated with 100 μL of 10 mM cysteine at 37°C for 4 hours. Following the removal of cysteine, the plate was washed twice with PBS, and 200 μL of tumor cells were seeded into each well. Once the target cells reached a plateau phase of growth, CAR-T cells were added at an E:T ratio of 1:1. Data collection continued until the end of the experiment.

PBMCs were isolated from three different healthy donors, irradiated with 20 Gy, and then combined. To distinguish between CAR-T cells and allogeneic PBMCs (allo-PBMCs) from the donors, the allo-PBMCs were labeled using the CellTrace CFSE Cell Proliferation Assay Kit (Thermo Fisher Scientific, C34570). CAR-T cells were co-cultured with allo-PBMCs at an E:T ratio of 1:1 for 24 hours. The expression of the activation marker CD69 on CAR-T cells was measured using flow cytometry analysis, while the apoptosis of allo-PBMCs was assessed using an apoptosis assay kit (Vazyme, A213-01).

PBMCs were isolated from three different healthy donors, and CD4⁺ and CD8⁺ T cells were separated using CD4 microbeads (Miltenyi, 130-097-048) and CD8 microbeads (Miltenyi, 130-045-201). The CAR-T cells were irradiated and labeled with CFSE. CD4⁺ and CD8⁺ T cells were co-cultured with various CAR-T cells separately at a 1:1 E:T ratio for 24 hours. The expression of CD69 on host T cells and the apoptosis of CAR-T cells were subsequently examined.

Cells were washed with 1 ml of pre-chilled PBS and gently resuspended in 100 μL of binding buffer. Annexin V-PE and 7-AAD staining solutions were then added to the suspension, followed by incubation in the dark at room temperature for 10 minutes. Data were collected using flow cytometry.

NK cells were labeled with CellTrace CFSE and co-cultured with CAR-T cells at a 1:1 ratio in 200 μL of medium containing an anti-CD107a antibody. After 1 hour of co-culture, monensin was added to the medium at a concentration of 100 μg/mL, and the cell mixture was incubated for an additional 4 hours. After washing, the cells were resuspended in PBS and analyzed by flow cytometry. Degranulated NK cells were identified as the CFSE⁺ CD107a⁺ population.

Macrophages and CAR-T cells were labeled with CellTrace CFSE and CellTrace Far Red, respectively. Macrophages were then co-cultured with target cells at a 1:1 ratio in 200 μL of medium for 6 hours. After washing, the cells were resuspended in PBS for flow cytometry analysis. Macrophages that phagocytized target cells were identified as the population double-positive for both fluorescent dyes.

Protein samples were prepared by lysing cells with RIPA lysis buffer. The lysates were then boiled and denatured before separation by SDS-PAGE. Proteins were subsequently transferred onto a polyvinylidene difluoride (PVDF) membrane and incubated overnight with primary antibodies from Cell Signaling Technology, including phospho-STAT5 (Tyr694; #9359), phospho-BCL2 (Ser70; #2827), and β-Actin (#93473). After washing, the membrane was incubated with a secondary antibody for 1 hour, and protein expression levels were detected using autoradiography.

Cell supernatants were collected during cytotoxicity detection assay and the levels of IL-6, IL-10, TNF-α, IFN-γ and Granzyme B secreted by CAR-T cells were assessed using the BD™ Cytometric Bead Array (CBA) Kit according to the manufacturer’s protocol.

Animal experiments were conducted in accordance with the regulations of the Animal Experiment Ethics Committee of Nanjing Normal University (IRB number: 2020-0047). To construct humanized immune system mice, 1×10⁷ PBMCs were injected intravenously via the tail vein into C-NKG mice (Cyagen, China) in a volume of 200 μL. Starting from the day of infusion, blood was collected from the tail vein every 7 days, and flow cytometry was used to determine the proportion of CD45⁺ cells, assessing the number of CAR-T cells.

For the xenograft models, 5×10⁶ U251 or 786-0 cells were subcutaneously injected into 6-week-old female C-NKG mice. When tumor volumes reached approximately 100 mm³, treatment was administered by intravenously injecting the indicated T cells in a volume of 200 μL. Tumor size, body weight, skin condition, fur quality, and posture of each mouse were continuously monitored. Mice were euthanized upon meeting predefined euthanasia criteria (significant weight loss, signs of distress) or as recommended by the veterinary staff. Tumors and other tissues, including heart, liver, spleen, lungs, kidney and brain were extracted.

Blood samples were collected 8 days post infusion. Genomic DNA was isolated from these samples using the DNeasy Blood & Tissue Kit (QIAGEN, 69506) following the manufacturer’s protocol. CD70 CAR copy numbers in genomic DNA were quantified via qPCR using the primers listed in Supplementary Table S1 . To measure cytokine levels, mouse blood was left at room temperature for 20 minutes, centrifuged to obtain serum, and levels of IL-6, TNF-α, and IFN-γ were analyzed using the CBA assay.

Mouse tissues were fixed with 4% paraformaldehyde (Boston BioProducts), embedded in paraffin, sectioned, and stained with H&E or indicated antibodies for IHC experiments. Images were obtained with a 3DHISTECH Panoramic digital slide scanner and the associated CaseViewer software (3DHISTECH).

PBMCs from the same donor were used to prepare CAR-T, SAP UCAR-T, and CD47 UCAR-T simultaneously. Cell pellets were collected and resuspended in TRIzol. RNA sequencing and subsequent analyses were performed by Genepioneer (Nanjing, China).

All experiments were performed independently at least three times. Experimental data were analyzed and plotted using GraphPad Prism 8.0 software and are presented as means ± SD. The Student’s t-test was employed to compare differences between two groups, while one-way and two-way analysis of variance (ANOVA) were used for statistical analysis among multiple groups. Differences were considered statistically significant when p < 0.05: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant.