Engineered CD70 CAR-T cells boosted by TS-2021 show strong tumor-fighting effects against brain cancer

The U251 cell line was obtained from the Cell Center, Chinese Academy of Medical Sciences (Beijing, China), the 293T cell line was obtained from ATCC, and the human astrocytes (HA) were obtained from ScienCell Research Laboratories. BT325 and BT01 are primary GBM cells that were developed at the Beijing Neurosurgical Institute [36, 37]. The U251-Luc, BT325-Luc, and BT01-Luc cell lines were established through transduction with lentiviral vectors encoding the luciferase gene (Luc). GBM cells and HEK293T cells were maintained in DMEM (Gibco, USA) supplemented with 10% fetal bovine serum (Gibco, USA) and 1% penicillin/streptomycin (Gibco, USA). HA were cultivated in Astrocyte Medium (AM, 1801, ScienCell). The cell lines used in this study were mycoplasma-free.

The CD70 CAR (consisting of the CD8 signal peptide, trCD27, CD8 hinge & transmembrane domain, 4-1BB, CD3ζ, and EGFP) was engineered following a published method [35], with the CD8 signal peptide inserted before trCD27 and an EGFP reporter used after the self-cleaving peptide sequence. The Mock CAR was designed using the above method. Then, the CAR sequences were synthesized and inserted into a second-generation lentiviral vector (GeneChem Co., Ltd., China) under the control of the human EF1a promoter.

First, 293T cells were expanded in DMEM supplemented with 10% fetal bovine serum. A recombinant lentivirus was prepared through plasmid transfection (CAR vector plasmid, pMD2. G, and psPAX2) into 293T cells. The supernatant was collected 72 h post-transfection, and the lentivirus was filtered through sterile Millex filters (Millipore), concentrated with Amicon Ultra filters (Millipore), and preserved at − 80 °C.

Isolation of human T cells and generation of CAR-T cells.

Human peripheral blood mononuclear cells (PBMCs) were isolated from the blood of healthy donors using a lymphocyte separation medium (LBS1077, Precision BioMedicals, China). Human CD3⁺ T cells were isolated and purified from PBMCs via immunomagnetic negative selection (17951, STEMCELL) and then activated for 48 h with the soluble ImmunoCult™ Human CD3/CD28/CD2 T-Cell Activator (STEMCELL) and 200 U/mL IL2. Activated T cells were transfected with CD70 or Mock CAR-encoding lentivirus (MOI = 10:1) and 5 µg/mL polybrene, followed by centrifugation at 2100 rpm for 90 min at room temperature. The cells were cultured in X-vivo 15 medium (Lonza) supplemented with 200 U/mL IL2 for six days. The CAR-positive rate exceeded 50%.

The construction of OAd-GFP (wild-type human adenovirus type 5 genome lacking the E1B19K and E3 regions, with the human Ki67 promoter linked to the human TGF-β2 5′UTR replacing the E1A promoter and the EGFP gene inserted at the E3 region) and TS-2021 (wild-type human adenovirus type 5 genome with deletions in the E1B19K and E3 regions, the human Ki67 promoter linked to the human TGF-β2 5′UTR replacing the E1A promoter, and the human IL15 gene replacing the EGFP) was described previously [29].

First, OAd was added to T/CAR-T cells at an MOI of 400:1 in the presence of 1× HitransG P (GeneChem Co., Ltd., China), followed by centrifugation at 2100 rpm for 90 min at room temperature. T/CAR-T cells were examined 48 h after infection with OAd-GFP via fluorescence microscopy. Infection rates were assessed at 48 and 144 h post-infection via flow cytometry (FCM).

To evaluate the infectivity of OAd-GFP, OAd-GFP (1.2 × 10⁷VP), CD70 CAR-T^OAd−GFP (2 × 10⁴), or Mock CAR-T^OAd−GFP (2 × 10⁴) was added to U251 cells (4 × 10⁴) or HA (4 × 10⁴). The fluorescence intensity of GFP was measured using an Axio Observer inverted microscope (ZEISS, Germany), and the data were analyzed using the ImageJ software (NIH, USA).

T/CAR-T cells (5 × 10⁴) were plated in 12-well plates containing X-vivo 15 medium (Lonza) enriched with 200 U/mL IL2 after OAd was added. These T/CAR-T cells were counted every day. To test the viability of T/CAR-T-cells, the above cells were treated with 0.4% trypan blue for 3 min at room temperature and then analyzed using a hemocytometer. The number of stained cells and the total number of cells were calculated to determine the cell viability.

CAR-T^OAd−GFP and CAR-T^TS−2021 cells were lysed by two freeze-thaw cycles, and the resulting lysates containing OAd were collected. Viral DNA was separated from the cell lysates using the TIANamp Virus DNA/RNA Kit (Tiangen Biotech, Beijing, China). The virus titer was measured following the method described in another study [38], and the E1A cycle threshold (Ct) in the viral DNA was quantified via E1A-specific quantitative polymerase chain reaction (qPCR). The viral titer was quantified by applying the Ct values to the E1A standard curve (Fig. s8) established using OAd (5.0 × 10¹¹VP/mL). The E1A sequences of primers used were as follows: 5′-AACCAGTTGCCGTGAGAGTTG-3′ (forward) and 5′-TCGTTAAGCAAGTCCTCCTCGATACAT-3′ (reverse).

Cytotoxicity was evaluated by co-culturing CAR-T/CAR-T^OAd cells with GBM-Luc target cells (2 × 10⁴/well) at an E: T ratio of 1:1 in 24-well plates. GBM cells (2 × 10⁴/well) were treated with OAd-GFP or TS-2021 (MOI = 300:1). Next, 2 µL of 15 mg/mL D-luciferin (LABLEAD, China) was added to each well after 24, 48, and 72 h of treatment. Bioluminescence was measured using the IVIS Spectrum system (PerkinElmer) and analyzed using Living Image 4.4 (PerkinElmer) to assess cytotoxicity.

In some studies, to investigate the cytotoxicity mediated by CAR-T cells under cytokine stimulation, IL2 (1.2 ng/mL, STEMCELL), IL7 (0.1 ng/mL, STEMCELL), IL12 (2 ng/mL, STEMCELL), and IL15 (0.2 ng/mL, STEMCELL) were added to the culture medium at their effective dose 50 (ED50) concentrations while co-culturing CAR-T cells with target cells.

Flow cytometry analysis was conducted using a BD Accuri C6 Plus instrument (BD Biosciences). The positive expression rate of CAR and the viral infection rate in T cells were determined by detecting EGFP fluorescence signals. Fluorochrome-conjugated antibodies were used for staining following the manufacturer’s protocol. The antibodies used are detailed in Supplemental Table. 1

CAR-T, CAR-T^OAd−GFP, and CAR-T^TS−2021 cells were incubated with U251 cells (10⁵/well) in 12-well plates (E: T = 1:1) for 48 h, and the cell-free supernatant collected was analyzed using a Human Cytokine Array Q2000 kit (QAH-CAA-2000, RayBiotech) following the manufacturer’s instructions.

After treating GBM cells (10⁵ cells per well in six-well plates) with OAd at an MOI of 300:1 for 72 h, the supernatants were collected. Next, CAR-T cells (5 × 10⁴) were plated in 12-well plates containing X-vivo 15 medium (Lonza) enriched with 200 U/mL IL2 after adding OAd, and the cell-free supernatants were harvested at 48 h, 96 h, and 144 h. The level of IL15 secretion was quantified in the supernatants collected from GBM cells or CAR-T cells using a human IL15 ELISA kit (EHC013, NeoBioscience) following the manufacturer’s guidelines. After OAd was added, CAR-T cells (5 × 10⁴) were plated in 12-well plates containing X-vivo 15 medium (Lonza) without IL2. The supernatants were collected after 72 h, and IL2 and IL15 secretions were measured using a human IL2 ELISA kit (EHC003, NeoBioscience) and the aforementioned human IL15 ELISA kit following the manufacturer’s instructions.

Two sgRNAs were designed that target a common sequence present in all CD70 transcripts to delete a large fragment of the CD70 coding sequence (CDS). The sgRNA sequences used were as follows: KO#1: 5′-CAGCAGGCTGATGCTACGGG-3′ and KO#2: 5′-CCATGTAGATGCCATCACGA-3′. A lentiviral vector (pLenti-Cas9-CD70 KO) was constructed, and the lentivirus was produced by Shanghai GeneChem Co., Ltd. A flow cytometry assay was conducted to analyze CD70 expression in BT325 and U251 cells seven days after transduction with the CD70 KO-encoding lentivirus.

CAR-T, CAR-T^OAd−GFP, and CAR-T^TS−2021 cells were collected three days after adding OAd. Total RNA was extracted, followed by PCR amplification, cDNA library preparation, and subsequent bioinformatics analysis. These processes were conducted by Suzhou GENEWIZ Biotech Co., Ltd.

Total RNA was extracted from HA, GBM, and CAR-T cells using Total RNA Isolation Reagent (BS258A, Biosharp, China) following the manufacturer’s guidelines. Next, cDNA was synthesized using All-in-One First-Strand Synthesis MasterMix (F0202, LABLEAD, China). Gene expression was quantified using specific primers and analyzed via SYBR qPCR (Applied Biosystems), with β-actin as the internal control. The sequences of primers used were as follows: 5′-CTGGCTGCTGACCGAGG-3′ (β-actin-F), 5′-GAAGGTCTCAAACATGATCTGGGT-3′ (β-actin-R), 5′-CCATATGCCTGTGGAGTGGAA-3′ (Ki67-F), and 5′-CCACCCTTAGCGTGCTCTTGA-3′ (Ki67-R).

Protein samples were isolated from cells using RIPA lysis buffer (R0020, Solarbio, China) containing a protease inhibitor (Roche) and a phosphatase inhibitor (Roche). The cell lysates were then centrifuged at 10,000×g at 4 °C, and the supernatants were collected to determine the protein concentration using the bicinchoninic acid (BCA) method. Proteins were loaded onto gels, separated by SDS-PAGE, and then transferred to polyvinylidene fluoride membranes. The membranes were blocked at room temperature with TBST (Tris-buffered saline, 0.1% Tween 20) containing 5% non-fat dry milk for 1 h. After the membranes were blocked, they were incubated overnight with primary antibodies at 4 °C, washed with TBST, and incubated with secondary antibodies at room temperature for 60 min. The protein bands were visualized using an enhanced chemiluminescence kit (NCM Biotech, China) and imaged using a ChemiDoc MP system (Bio-Rad, USA). The antibodies used are detailed in Supplemental Table 1.

The NCG mice were obtained from GemPharmatech Co., Ltd. and housed in a specific pathogen-free environment. Under isoflurane anesthesia, GBM cells were stereotaxically injected into the right striatum (using the bregma as the reference point; 2 mm lateral, 1 mm anterior, and 3 mm deep) to establish an orthotopic GBM model. The mice were engrafted with GBM cell lines as described in the individual experiments. These mice were then sorted before treatment allocation to ensure that the mean Flux values were similar across all treatment groups. OAd or CAR-T cells were intratumorally injected at the indicated time points. Tumor growth was monitored in vivo via bioluminescence imaging using an IVIS Spectrum (PerkinElmer) system after intraperitoneal injection of D-luciferin substrate solution. In some experiments, blood samples were obtained from mice via postorbital venous puncture 10 days after treatment. Cytokine levels were measured using the Mouse Inflammation Array Kit (FAM-INF-1-48, RayBiotech) and the Human IL15 ELISA Kit (EHC013, NeoBioscience) following the manufacturer’s guidelines. The body weight and survival status of the mice were monitored regularly. Bioluminescence signals were quantified using Living Image 4.4 (PerkinElmer).

The differences between or among groups were determined by Student’s t-test or ANOVA. Survival curves were analyzed by conducting the log-rank (Mantel-Cox) test. Statistical significance was set at a p < 0.05. The data are presented as the mean ± SD, and all analyses and visualizations were conducted using the GraphPad software (version 9.4).

Although OAd-GFP, a genetically engineered third-generation OAd (Fig. S1A), selectively infects and lyses GBM cells [29], whether non-tumor cells, like tumor-targeted CAR-T cells, can be infected by OAd-GFP remains uncertain. Human CD70 or Mock CAR-T cells, with about 55% CAR-positive cells (Figs. 1A, S1B, S1C, and S1D), were generated and cultured for six days. Adenoviruses enter cells via interactions between their fiber proteins and coxsackie-adenovirus receptor (CXADR) on the cell surface [39]. The results of the flow cytometry analysis revealed that CXADR expression was lower in CAR-T cells than in GBM cells and HA (Fig. S1E). However, the results of qPCR assays revealed that Ki67 mRNA was highly expressed in CAR-T cells (Fig. S1F), and the results of Western blotting assays indicated stable expression of the TGF-β2 protein in CAR-T cells, although its expression level was lower in these cells than in GBM cells (Fig. S1G). These results indicated that once OAd-GFP infects CAR-T cells, it can replicate within them.

As shown in Fig. S2A, after eight days of expansion of activated T cells, directly adding OAd-GFP to T cells at multiplicities of infection (MOIs) of 100:1, 200:1, 400:1, or 800:1 without spin infection for 48 h did not lead to significant OAd-GFP infection in T cells. To infect T cells with OAd-GFP, we followed a protocol resembling the lentiviral transduction method used in CAR-T-cell production, utilizing HitransG P (a less toxic alternative to polybrene) and performing centrifugation at 2100 rpm for 90 min. The OAd-GFP infection rate in T cells increased to 30% at an MOI of 400:1 via this alternative method (Fig. S2A). Although the addition of HitransG P and centrifugation for 90 min did not affect T-cell proliferation significantly, OAd-GFP-infected T (T^OAd−GFP) cells showed reduced proliferation (Fig. S2B) and progressively declining viability (Fig. S2C) about six days after OAd-GFP infection at an MOI of 800:1. In contrast, at an MOI of 400:1, OAd-GFP had a limited effect on T-cell proliferation and viability (Figs. S2B, S2C, and S2D), with the OAd-GFP positivity rate in T cells remaining at 17.8% even seven days post-infection (Fig. S2E). We added OAd-GFP to CD70 CAR-T cells that had expanded for six days at an MOI of 400:1, along with HitransG P, and centrifuged them for 90 min (Fig. 1A). The proliferation and viability of CD70 CAR-T cells infected with OAd-GFP (CAR-T^OAd−GFP) did not decrease significantly at six days post-infection (Fig. 1B and C, and 1D). These findings showed that there is a seven-day window during which CAR-T^OAd−GFP cells can deliver OAd-GFP into GBM cells.

We next examined whether CAR-T^OAd−GFP cells could deliver OAd-GFP to GBM cells and whether this process depended on the tumor antigen specificity of CAR-T cells. For CD70⁺ U251 cells (Fig. S3A), CD70 CAR-T^OAd−GFP cells, at an effector-to-target (E: T) ratio of 1:2, effectively delivered OAd-GFP into the cells, unlike Mock CAR-T^OAd−GFP cells (Fig. 1E). A Transwell assay was conducted to exclude trogocytosis [40] as a potential mechanism for transferring the GFP gene. GFP⁺ U251 cells were observed in the bottom compartment (Fig. 1F), probably because CD70 CAR-T^OAd−GFP cells, rather than Mock CAR-T^OAd−GFP cells, were reactivated by U251 cells in the top compartment, leading to the release of OAd-GFP and subsequent diffusion to the U251 cells below. However, adding CD70 CAR-T^OAd−GFP cells to CD70^– HA (Fig. S3A) did not cause infection of U251 cells in the bottom compartment (Fig. 1G). These results highlighted that CAR-T^OAd−GFP cells efficiently delivered OAd-GFP to GBM cells in a target antigen-dependent manner.

We subsequently evaluated the antitumor activity of OAd-GFP-infected CAR-T cells by conducting in vitro cytotoxicity assays. An E: T ratio of 1:1 was used for CAR-T cells and GBM cells. First, the cytotoxic activity of CD70 CAR-T cells against various GBM cell lines was evaluated. The expression of CD70 was detected in GBM cells but was not detected in HA (Fig. S3A). The percentage of CD70⁺ cells followed the order BT325 = U251 > BT01, whereas the intensity of CD70 expression was the highest in BT325, followed by that in U251 and BT01 (Figs. S3B and S3C). These differences highlighted the antigenic heterogeneity of GBM cells. After co-culturing CAR-T cells with GBM cells for 24 h, CD70 CAR-T cells showed the strongest cytotoxicity against BT325 cells, while the cytotoxicity against CD70-low-expressing BT01 cells was low (Fig. S3D). The results of a simple linear regression analysis suggested that CD70 CAR-T-cell-mediated killing may depend on the density of CD70 targets and the percentage of CD70⁺ cells. (Figs. S3E and S3F). Increasing the E: T ratio to 4:1 improved the cytotoxicity of CD70 CAR-T cells to target cells (Fig. S3G). After adding CD70 CAR-T^OAd−GFP cells to GBM cells, the antitumor efficacy against GBM cells progressively increased over 72 h and significantly surpassed that of CD70 CAR-T cells following 72 h of co-culture (Fig. 2A). In BT01 cells, CD70 CAR-T^OAd−GFP significantly enhanced the in vitro antitumor efficacy compared to that of CD70 CAR-T cells (Fig. 2A).

To investigate the ability of CD70 CAR-T^OAd−GFP cells to eradicate pre-established tumors in vivo, an NCG mouse tumor burden model was constructed (intracranial orthotopic injection of U251 cells, with control group mice having an approximate survival time of 40 days after the initiation of treatment), followed by intratumoral injection of CD70 CAR-T^OAd−GFP (Fig. 2B). Although intratumoral injection of OAd-GFP or CD70 CAR-T cells suppressed tumor progression and prolonged mouse survival, the CAR-T^OAd−GFP combination had a strong antitumor effect, leading to tumor regression and long-term survival (Fig. 2C and D, and 2E). These results indicated that CAR-T^OAd−GFP cells exhibit robust anti-GBM efficacy and can overcome CAR-T-cell therapy resistance caused by antigen heterogeneity.

The cytotoxicity of CD70 CAR-T cells against GBM cells did not significantly increase with prolonged co-culture time (Fig. 2A). We hypothesized that this may be due to changes in the state of CAR-T cells during their co-culture with target cells. Additionally, as OAd-GFP carrier cells, these changes might decrease the cytotoxic effectiveness of CAR-T^OAd−GFP. The expression levels of exhaustion markers in CD70 CAR-T cells were assessed following their co-culture with CD70⁺ GBM cells. T-cell exhaustion markers were significantly upregulated after 72 h of antigen stimulation, whereas their upregulation remained relatively limited after 24 h of co-culture (Fig. S4A). When co-cultured with CD70– GBM cells (Fig. S4B) instead of with CD70⁺ GBM cells for 72 h, T-cell exhaustion markers were not significantly upregulated (Fig. S4C). When CD70 CAR-T cells were co-cultured with CD70⁺ GBM cells and supplemented with IL2, IL7, IL12, and IL15, we found that IL7 and IL12 did not increase CD70 CAR-T-cell cytotoxicity (Fig. S5A). IL2 promoted the cytolysis of CD70 CAR-T cells, although the difference was not significant (Fig. S5A). In contrast, IL15 enhanced the cytotoxicity of CD70 CAR-T cells (but not Mock CAR-T cells) and decreased the level of expression of T-cell exhaustion markers under antigen stimulation (Figs. S5A, S5B, and S5C). These results indicated that IL15 can enhance the cytotoxic effect of GBM-specific CAR-T cells and alleviate CAR-T-cell exhaustion under continuous antigen stimulation.

TS-2021 is a modified version of OAd-GFP in which IL15 replaces GFP (Fig. 3A). When GBM cells are infected, TS-2021 facilitates the synthesis and release of IL15 by GBM cells (Figs. S6A and S6B). CD70 CAR-T cells infected with TS-2021 (using the same transfection method as OAd-GFP) showed higher proliferative capacity and viability within six days post-infection (Fig. 3B and C). IL15 secretion was detectable in the supernatant starting from day 2 post-infection (Fig. 3D). On day 5 post-infection, flow cytometry was performed to evaluate the differentiation status of various CAR-T cells. Compared to CAR-T cells, CAR-T^TS−2021 had a greater proportion of naive/stem cell memory T cells (CD45RA⁺CCR7⁺) [41], whereas CAR-T^OAd−GFP did not exhibit this characteristic (Fig. 3E). After 72 h of co-culture with CD70⁺ U251 cells, CAR-T^OAd−GFP cells presented an exhaustion phenotype similar to that of CAR-T cells, whereas CAR-T^TS−2021 showed a lower expression of exhaustion markers (Fig. 3F). The OAd load in CAR-T cells was dynamically assessed post-infection via qPCR, and the results showed that the average OAd load in CAR-T^TS−2021 cells increased after five days (Fig. 3G). When different CD70 CAR-T cells were incubated with GBM cells (E: T = 1:1) for 72 h, CAR-T^TS−2021 exhibited superior cytolytic activity compared to CAR-T^OAd−GFP (Fig. 3H). In vitro cytokine production was assessed in CAR-T, CAR-T^OAd−GFP, and CAR-T^TS−2021 cells co-cultured with CD70⁺ U251 cells. In OAd-infected CAR-T cells, IFNγ and IL2 levels increased, whereas IL10 and GM-CSF levels decreased. IL15 levels increased only in TS-2021-infected CAR-T cells. The secretion levels of TNFα, IL4, and IL6 were not significantly different among the three cell types (Fig. 3I). Unlike CAR-T cells, OAd-infected CAR-T cells did not induce excessive cytokine release under in vitro conditions. These findings indicated that CAR-T^TS−2021 cells showed greater proliferation, persistence, and cytotoxicity in vitro than CAR-T^OAd−GFP cells. Additionally, OAd-infected CAR-T cells did not induce significant cytokine release toxicity in vitro.

We subsequently evaluated the in vivo antitumor efficacy of CAR-T^TS−2021 cells. Based on the strong performance of CD70 CAR-T^OAd−GFP cells in the NCG mouse tumor burden model, we established a high tumor burden model by intracranially injecting U251 cells into NCG mice (the mice in the control group had a survival time of about 20 days post-treatment). CD70 CAR-T^TS−2021 cells were then administered via intratumoral injection, and blood samples were obtained via postorbital venous puncture 10 days after treatment to measure cytokine levels (Fig. 4A). In the high-tumor burden model, TS-2021 showed low efficacy (Fig. 4B). CD70 CAR-T-cell therapy delayed tumor growth and extended survival, probably due to the high expression of CD70 in U251 cells. Both CD70 CAR-T^OAd−GFP and CD70 CAR-T^TS−2021 intratumoral injections strongly inhibited tumor growth and preserved mouse body weight compared to CD70 CAR-T-cell injections. Compared to that in the CAR-T^OAd−GFP group, the survival time of the mice in the CAR-T^TS−2021 group was significantly longer (Fig. 4C and D, and 4E). In contrast to the intratumoral injection of CD70 CAR-T cells into mice, CD70 CAR-T^TS−2021 treatment did not increase serum cytokine levels. Mice treated with CD70 CAR-T^TS−2021 cells showed reduced cytokine levels in the serum compared to those treated with CAR-T^OAd−GFP cells (Fig. 4F). Additionally, human IL15 (hIL15), which is essential for sustaining CAR-T-cell activity, was identified in the peripheral blood of mice in the CD70 CAR-T^TS−2021 group (Fig. 4G). These results indicated that CAR-T^TS−2021 cells exhibited stronger in vivo anti-GBM efficacy than CAR-T^OAd−GFP cells, while CAR-T cells infected with TS-2021 did not induce increased cytokine release toxicity.

To better understand how TS-2021 infection enhances the antitumor efficacy of CAR-T cells, we conducted RNA sequencing on CD70 CAR-T, CD70 CAR-T^OAd−GFP, and CD70 CAR-T^TS−2021 cells. A clustering heat map was made to illustrate the differences in gene expression among various CAR-T cells (Fig. 5A). The differential gene volcano plot revealed that OAd-GFP, but not TS-2021, had a stronger effect on the changes in gene expression in CAR-T cells (Fig. 5B). GO analysis was performed on the genes with differential expression, highlighting the top 16 most significantly enriched GO terms. In CAR-T^TS−2021 cells, the term “positive regulation of cell proliferation” was prominent, whereas in CAR-T^OAd−GFP cells, the terms “cellular response to DNA damage stimulus” and “double-strand break repair” were prominent (Fig. S7). DNA damage-related genes were primarily upregulated in CAR-T^OAd−GFP, whereas IL15 was upregulated only in CAR-T^TS−2021. IL2 was expressed in both CAR-T^OAd−GFP and CAR-T^TS−2021 (Fig. 5C). The results of the immunoblotting analysis revealed that γ-H2AX, a DNA damage marker for double-strand breaks [42], was highly expressed in CAR-T^OAd−GFP cells, whereas cleaved caspase-3, a marker of apoptosis [43], was expressed at the lowest level in CAR-T^TS−2021 cells (Fig. 5D). The expression of poly-ADP-ribose (PAR) in CAR-T^OAd−GFP and CAR-T^TS−2021 cells suggested that OAd entry into CAR-T cells activated the poly (ADP-ribose) polymerase (PARP) repair function. Additionally, the higher expression of PARP in CAR-T^TS−2021 cells indicated increased repair efficiency, which was consistent with the increased expression of PAR in these cells. PARP expression was downregulated in CAR-T and CAR-T^OAd−GFP cells because of cleavage by cleaved caspase-3 (Fig. 5D). Consistent with the RNA sequencing results, IL15 synthesis increased only in CAR-T^TS−2021 cells, whereas IL2 expression was upregulated in CAR-T^OAd−GFP and CAR-T^TS−2021 cells, as determined by cytokine assays of the supernatant (Fig. 5E). The level of expression of γ-H2AX in CAR-T^OAd−GFP cells was lower after 72 h of IL15 supplementation than in cells without IL15 treatment (Fig. 5F). These results indicated that OAd infection induces DNA damage in CAR-T cells, whereas IL15 effectively alleviates this damage.

To further assess how IL15 mitigates OAd-induced DNA damage in CAR-T cells, we conducted the Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of differentially expressed genes and highlighted the 16 most significantly enriched pathways (Fig. 6A). Pathways involved in T-cell proliferation have received significant attention. In this study, the JAK-STAT and MAPK pathways were enriched in CAR-T^TS−2021 cells. In contrast, CAR-T^OAd−GFP was enriched primarily in pathways associated with viral infection. The JAK-STAT pathway was also enriched, probably due to the synthesis and release of IL2 induced by viral infection. Next, we confirmed the results of the differential gene KEGG enrichment analysis by conducting Western blotting assays (Fig. 6B). Both p-JAK1 and p-STAT3 were upregulated in CAR-T^OAd−GFP and CAR-T^TS−2021 cells, suggesting that the JAK-STAT pathway was activated in CAR-T cells following OAd infection. In contrast, p-STAT5, p-MEK, and p-ERK were significantly expressed only in CAR-T^TS−2021, probably due to the synthesis and release of IL15. IL15 was added to the culture medium of CAR-T^OAd−GFP cells. The results revealed that CAR-T^OAd−GFP enhanced proliferation and viability over six days post-infection, although it lagged behind that of CAR-T^TS−2021 cells (Fig. 6C and D). The advantage in CAR-T^TS−2021 cells may be attributed to the continuous release of IL15 in these cells. In the presence of IL15, the levels of p-STAT5, p-MEK, and p-ERK increased in CAR-T^OAd−GFP cells (Fig. 6E). These results suggested that IL15 promotes the proliferation and viability of CAR-T^TS−2021 cells by activating the JAK-STAT and MAPK-ERK signaling pathways, thus reversing OAd-induced DNA damage in CAR-T cells.

GBM is highly heterogeneous, and mutations or loss of its target antigens can greatly limit the efficacy of CAR-T-cell therapy [44]. GBM cell models were constructed by mixing CD70⁺ and CD70– GBM cells in different proportions (80%, 50%, and 20% CD70⁺) in vitro (Fig. 7A and B) to mimic antigen mutations or loss (AML) in GBM. As the CD70 positivity rate in GBM cells decreased, the cytolytic capacity of CAR-T cells also decreased. At 50% CD70 positivity, OAd-infected CAR-T cells maintained strong antitumor activity, with CAR-T^TS−2021 demonstrating greater efficacy over CAR-T^OAd−GFP. However, when the CD70 positivity rate decreased to 20%, the cell-killing ability of all CAR-T cells decreased considerably (Fig. 7C and D). When the E: T ratio increased to 4:1, OAd-infected CAR-T cells exhibited significantly stronger antitumor activity, primarily due to the OAd present in the cells. We intracranially implanted 80% CD70⁺ and 50% CD70⁺ U251 cells to construct a high tumor burden model. NCG mice treated with CAR-T^TS−2021 survived significantly longer than those in the CAR-T and CAR-T^OAd−GFP groups. However, when targeting tumors with 50% antigen positivity, the in vivo tumor-targeting effectiveness of CAR-T^TS−2021 decreased. These results suggested that CAR-T^TS−2021 can effectively inhibit GBM growth in AML in vitro and in vivo, partially overcoming AML-associated acquired resistance in GBM.