The different cancer cell lines including SPC-A-1 (lung adenocarcinoma), MDA-MB-231 and SKBR-3 (breast cancer), which were obtained from the Cell Bank of Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences (Shanghai, China) and Kunming Cell Bank of Type Culture Collection (Kunming Institute of Zoology, Kunming, China), were maintained in RPMI1640 medium (Biological Industries, Cat#01-100-1ACS, Israel) containing 10% FBS (Biological Industries, Cat#04-001-1ACS, Israel) and 1% penicillin/streptomycin (Biological Industries, Cat#03-031-1B, Israel) at 37°C in a humidified 5% CO₂ atmosphere. Mycoplasma contamination in cell cultures was detected by Nested PCR using the Mycoplasma PCR Detection Kit (Beyotime Biotechnology, Cat#C0301S, Shanghai, China) following the supplier’s recommended protocol.

Peripheral blood mononuclear cells (PBMCs) were harvested from the peripheral blood samples from healthy volunteers, all of whom tested negative for HBV, HCV, HIV, and syphilis in pre-transfusion infectious disease screening. An ethics committee of Kunming Medical University, Yan’an Affiliated Hospital, approved this study (2017-074-01). We hereby declare that informed consent was obtained for the use of human samples in the experiment and that the privacy of the donors was always respected.

The PiggyBac DNA transposon and transposase system (PB) was employed to mediate gene transfer into donors’ freshly isolated T lymphocytes, thereby inducing the expression of human HER2-specific CAR. The CAR consists of a scFv targeting human HER2 (either anti-HER2 or anti-HER2-13) and is incorporated CD28, 4-1BB and CD3ζ intracellular signaling domains, connected through a CD28-derived transmembrane region and CD8α hinge domain (Fig. 1A). The cDNA encoding the HER2/HER2-13-specific chimeric antigen receptor (CAR) was custom synthesized by GenScript (Piscataway, NJ, USA). T2A and the puromycin resistance (PuroR) coding sequence were excised from pCDH-EF1-MCS-T2A-Puro lentiviral vector (System Bioscience, Catalog Number CD527A-1, San Francisco, CA, USA) using digestion, and anti-HER2/HER2-13 scFv coding sequence was directionally cloned into the PB-EF1-MCS-IRES-Neo piggyBac vector (System Biosciences, Catalog Number PB533A-2), restricted using Xba1 and Bgl2. The designed sites collectively replaced the MCS-IRES-neo fragment in the vector backbone.

The structure of the human HER2 antigen was first predicted using AlphaFold3. Subsequently, potential binding regions between the antigen and the two constructed scFv fragments were predicted using BepiPred-3.0 (https://services.healthtech.dtu.dk/services/BepiPred-3.0/ (accessed on 01 August 2025)). Finally, HADDOCK (High Ambiguity Driven Docking)-based molecular docking was employed to calculate the binding score between the antigen and humanized anti-HER2-13 or anti-HER2 scFv [23,24].

HER2 expression was knocked down (or overexpressed) in SKBR-3, MDA-MB-231, and SPC-A-1 cell lines through lentiviral vector-mediated transduction. A lentivirus-based shRNA expression vector designed to target HER2 was constructed by Shanghai Genechem Co., Ltd. (Shanghai, China). The gene-silencing oligonucleotide sequence specifically targeting the HER2 cDNA is shHER2-111366: GCAGTTACCAGTGCCAATATC; shHER2-111367: GCTCCAAGTGTTTGAGACTCT; shHER2-111368: GAACACAGCGGTGTGAGAAGT, and a non-targeting scrambled shRNA (shNC) served as the experimental control: TTCTCCGAACGTGTCACGT. SKBR-3 cells were transfected with each lentiviral construct at an optimal multiplicity of infection (MOI) of 50. Transfection efficiency was assessed by monitoring GFP-positive cells under a fluorescence microscope (Bio-Rad, 742BR1307, Hercules, CA, USA). The HER2 overexpression procedure in SPC-A-1 and MDA-MB-231 cells followed the same protocol as previously described.

Anti-HER2/HER2-13 CAR-T cells were transfected with PB using 2b-Nucleofector kit (Lonza, VVPA-1002, Allendale, NJ, USA) by an electroporator (Nucleofector 2b, 91720139, Lonza). The procedure was according to our previous studies, with minor modifications [12,13]. Briefly, peripheral blood-derived PBMC were transfected with either the PB-HER2 or HER2-13 CAR-puro plasmid in combination with the Super PiggyBac transposase. On the first day after electroporation, CD3/CD28 magnetic beads (ThermoFisher, 11131D, Waltham, USA, USA) was added to cells for stimulation at a ratio of 2 beads per cell, and the cytokines IL-2 (Peprotech, AF-200-07-250UG, Rocky Hill, NJ, USA) (300 U/mL), IL-7 (Peprotech, AF-200-02-1MG, USA) (10 ng/mL) and IL-21 (Peprotech, AF-200-21-1MG, USA) (30 ng/mL) purchased from Peprotech, New Jersey, USA, were also added into X-vivo 15 culture system (Lonza, 04-418, USA) at the same time. On the fifth day, the beads were removed from the CAR-T cells culture system. CAR-T cells were selected and enriched by applying gradient concentrations of puromycin in culture medium starting on day 12 post-transfection. The cytokine-containing medium was refreshed at 48 h intervals for a duration of 18–20 days (Fig. 1B).

CAR-T cells products underwent cryopreservation in NutriFreez® D10 Cryopreservation Medium (Biological Industries, 05-713-1A, Israel) at 2 × 10⁷ cells/mL in 1.8 mL chilled cryo-vials (Thermo Scientific, 377267, Lyngby, Denmark). The cryopreserved vials were placed in a Nalgene Cryo 1°C Freezing Container and preserved in a −80°C refrigerator. The cryopreserved vials were placed in long-term liquid nitrogen storage.

After one month or six months in liquid nitrogen, cryopreserved cells were rapidly thawed in a 37°C water bath with gentle agitation until minimal ice crystallization persisted, then gently resuspended in 3 mL of X-vivo 15 medium (pre-equilibrated to 37°C) within a 15 mL tube, pelleted by centrifugation (1000 rpm, 5 min) (Thermo, SN09060479, Waltham, MA, USA) for 5 min and washed with X-vivo 15 medium for living cell counting and further flow cytometry analysis.

CAR-T cell phenotypes were profiled by flow cytometry after the cells ex vivo expansion or thawing. In brief, 5 × 10⁶ cells were collected from each group of CAR-T cells, and cell pellets were obtained by centrifugation at 1000 rpm for 5 min, followed by careful aspiration of the supernatant, and 100 μL sheath fluid was used to resuspend the cells. Subsequently, the cell suspension was slightly mixed and incubated with the antibody at 4°C for 20 min in the dark. After incubation, cellular immunophenotyping was performed using flow cytometric analysis. The following antibodies were used in flow cytometry detection: anti-CD3 PerCP-Cy5.5 (BD Pharmingen, Clone SP34-2, San Jose, CA, USA), anti-CD4 PerCP-Cy5.5 (BD Pharmingen, Clone L200), anti-CD8 PerCP-Cy5.5 (BD Pharmingen, Clone RPA-T8), anti-Myc-Tag PE (CST, #3739), anti-CD44 PerCP-Cy5.5 (BD Pharmingen, Clone G44-26), anti-CD62L PE (BD Pharmingen), anti-Tim-3 PE (BD Pharmingen, Clone 7D3), anti-PD-1 PE (BD Pharmingen), anti-LAG-3 PE (BD Pharmingen, Clone T47-530). Antibodies were used at 1:51 dilutions (prepared by adding 1 μL antibody to 50 μL flow buffer supplemented with 1% BSA). The data were obtained from a BD FACSCanto and were analyzed using FlowJo version 10.5.3 (BD Biosciences).

Overexpression and knockdown of HER2 in tumor cells were determined by western blot. Following PBS washing (4°C), cells were lysed using ice-cold RIPA buffer (Beyotime Biotechnology, P0013, Shanghai, China) containing protease inhibitor mixtures (Roche Molecular Biochemicals, 4906845001, Basel, Switzerland). Protein content was assessed via the BCA Protein Concentration Assay (Beyotime Biotechnology, P0010), and 30 μg of total protein was loaded and electrophoresed on 10% SDS-PAGE gels, which were then transferred to a hydrophobic PVDF (Polyvinylidene Fluoride) membrane (Millipore, Carrigtwohill, ISEQ00010, Ireland). The membrane was blocked with 5% (w/v) non-fat dry milk in Tris-buffered saline + 0.1% Tween-20 (TBST, 9005-64-5, Sigma-Aldrich, USA) for about 1–2 h at room temperature. They were then incubated overnight at 4°C with monoclonal rabbit anti-human HER2 antibody (Cell Signaling Technology, 4290S, Danvers, MA, USA, 1:1000) or anti-β-actin mAb (mouse origin) (Proteintech, 66009-1, Rosemont, IL, USA, 1:1000). The membrane was then washed with TBST and incubated with the secondary HRP-goat anti-mouse IgG (H + L) (Proteintech, RGAM001, USA) or HRP-goat anti-rabbit IgG (H + L) (Proteintech, 10435-1-AP, USA) at room temperature for 1 h, with both antibodies at a dilution of 1:5000. Protein expression was visualized using ECL protein blotting substrate (Proteintech, PK10003, USA) and images were captured by using a BIO-RAD instrument (Bio-Rad, 732BR1872, USA).

The water-soluble tetrazolium salt (WST-8)-based CCK-8 assay (CCK-8, CK04-500T, DOJINDO, Kumamoto, Japan) was employed to quantify the tumoricidal effects of CAR-T cells. Briefly, tumor cells (5000 cells/each well) were seeded into 96-well plates overnight, then co-cultured with serially diluted CAR-T cells to quantify cytotoxicity over a range of E:T (effector cell/target cell) ratios. After 24 h, following removal of supernatant and non-adherent cells, the viability of residual tumor cells was quantified using CCK-8 reagent in accordance with the manufacturer’s instructions after washing the plate twice with PBS. Three replicate wells were set up for each experiment. The experiments were conducted in triplicate with three independently produced CAR-T cell batches.

Non-treated T cells (NT)/CAR-T cells were co-incubated with tumor cells at an effector-target ratio of 2:1 for 1 day. After that, the supernatant of the co-cultured cells was harvested by centrifugation (1000 rpm, 5 min), and clarified supernatants were aliquoted for subsequent cytokine profiling. The cytokines, including interleukin-2 (IL-2, EHC003.96), interleukin-6 (IL-6, EHC007.96), interleukin-10, (IL-10, EHC009.96), tumor necrosis factor-α (TNF-α, EHC103a.96), interferon-γ (IFN-γ, EHC102g.96), perforin (EHC154.96), granzyme B (EHC117.96) and granulocyte-macrophage stimulating factor (GM-CSF, EHC105.96) were detected by ELISA Kits in accordance with their instructions (Neobioscience Technology, Beijing, China).

To directly compare the efficacy of two groups of CAR-T cells in an in vivo setting, female NOD-scid mice (3–5 mice per group) were implanted with MDA-MB-231^HER2+ cells. The mice (6 weeks old, weighing around 18–20 g), supplied by SJA Laboratory Animal Co. (Hunan, China), were housed in pathogen-free facilities. And experimental procedures were approved by the Animal Experimentation Ethics Committee of Yan’an Hospital, Kunming Medical University (No. AEWC-2021100). Briefly, 5 × 10⁶ MDA-MB-231^HER2+ cells were transplanted into the breast fat pad of each mouse on day-10. On day 0, tumor growth was observed, and then the mammary cancer MDA-MB-231^HER2+ xenograft mice were randomly divided into control and treatment groups. The mice were injected with saline, trastuzumab (10 mg/kg) and CAR-T cells 2 × 10⁷ cells/100 μL by tail vein on days 0, 7 and 14, respectively. The mice in CAR-T groups were also administered with IL-2 (2000 IU/100 μL each mouse, Peprotech, AF-200-07-250UG, USA) intraperitoneally. Mouse body weights were tracked throughout the study period. At the end of the experiment, the mice were then sacrificed by cervical dislocation and the tumors were weighed and collected for immunofluorescence staining.

The collected tumor tissue samples from each group were subjected to paraffin embedding. After deparaffinization and antigen retrieval of the tumor tissue paraffin sections, apoptotic cells were detected in 5 μm deparaffinized tumor sections by using CF488 TUNEL Cell Apoptosis Detection Kit (Servicebio, G1504, Wuhan, China) according to the manufacturer’s instructions. Briefly, after sequential PBS rinses, the sections were subjected to enzymatic pretreatment with proteinase K (20 μg/mL) for 20 min at 37°C. Tissue sections were subjected to two 5-min PBS washes under gentle agitation, The slide was first incubated with 50 µL of Equilibration Buffer for 10 min at room temperature, and subsequently incubated with the TUNEL reaction mixture (2 µL recombinant terminal deoxynucleotidyl transferase enzyme, 5 µL CF488-conjugated dUTP, and 50 µL Equilibration Buffer) for 60 min at 37°C in a humidified dark chamber. After incubation, the slides were rinsed with PBS four times. For Ki-67 detection, slides were blocked with 5% BSA in a humidified chamber for 60 min, then incubated with anti-Ki-67 antibody (Proteintech, 27309-1-AP, USA, 1:200) overnight at 4°C. After three PBS washes, FITC-labeled goat anti-rabbit IgG (Thermo Fisher Scientific, 65-6111, Shanghai, China, 1:2000) was dispensed and incubated for 60 min in light-protected conditions. After both detections were washed with PBS, nuclear counterstaining was performed using DAPI (Thermo Fisher Scientific, D1306, China) under light-protected conditions for 10 min. Samples were coverslipped with a rapid-mounting media, and imaged using a fluorescence microscope (Bio-Rad, 742BR1307, USA). Green fluorescence from Ki-67 and TUNEL, respectively, indicated proliferating and apoptotic cells, while DAPI stains the cell nuclei.

To optimize transfection efficiency, electroporation trials were conducted using three different ratios of transposon to transposase plasmids for the generation of anti-HER2 and anti-HER2-13 CAR-T cells. Flow cytometric assessment showed that the proportion of CD3⁺CAR⁺ cells in anti-HER2 CAR-T populations was highest (53.8%) when a 5 μg:5 μg ratio of transposon to transposase plasmids was used, significantly exceeding the levels observed with 10 μg:5 μg and 10 μg:10 μg ratios (Fig. 1C). Moreover, the number of CD3⁺ CAR⁺ cells in anti-HER2 CAR-T populations was also higher in the group transfected with a 5 μg:5 μg ratio of the transposon to transposase plasmids, when compared with the other two ratio groups (Fig. 1E). Meanwhile, no significant differences in CD3⁺ CAR⁺ expression were observed among anti-HER-13 CAR-T cells transfected with the three different ratios of transposon to transposase plasmids (Fig. 1D). Notably, the anti-HER2-13 CAR-T group transfected with a 10 μg:5 μg ratio of transposon to transposase plasmids exhibited superior proliferation and increased prevalence of CD3⁺CAR⁺ cellular subsets relative to the other ratio groups (Fig. 1E).

As shown in Fig. 2A, we carried out a comparative analysis of the in vitro propagation of both CAR-T and NT cells. At days 7 and 14, the morphology of cells were observed under a microscope. The cells displayed a regular, round shape and tended to form clumps. The result demonstrated that CAR-T cells transfected with either of the two scFv sequences exhibited good viability and strong proliferative abilities over time, with no significant differences identified in both CAR-T cell groups (Fig. 2A,B).

To compare the CAR expression between anti-HER2 and anti-HER2-13 CAR-T cells, the transfected cells were cultured for approximately 16–20 days, and subsequently harvested for the detection of CD3⁺CAR⁺, CD4⁺CAR⁺, and CD8⁺CAR⁺ expression using flow cytometry. The percentages of CD3⁺CAR⁺, CD4⁺CAR⁺, and CD8⁺CAR⁺ expression in anti-HER2 CAR-T cells were 59.27 ± 11.84%, 15.11 ± 6.06%, and 24.98 ± 3.69%, respectively. In anti-HER2-13 CAR-T cells, these proportions were slightly elevated, reaching 64.04 ± 16.30% for CD3⁺CAR⁺, 15.82 ± 3.36% for CD4⁺CAR⁺, and 30.04 ± 3.08% for CD8⁺CAR⁺ cells (Fig. 2C–F). Moreover, the CD4⁺CAR⁺/CD8⁺CAR⁺ ratios of anti-HER2 and anti-HER2-13 CAR-T cells were also analyzed. The data revealed that the CD4⁺CAR⁺/CD8⁺CAR⁺ ratio of anti-HER2 CAR-T cells was 0.61 ± 0.24, while anti-HER2-13 CAR-T cells had a ratio of 0.53 ± 0.13. Nevertheless, no statistically significant difference was observed between the two CAR-T cell groups (Fig. 2G). Therefore, aside from the CD8⁺CAR⁺ population, the two CAR-T cell constructs exhibited minimal differences in cell expansion, as well as in the proportions of CD3⁺CAR⁺, CD4⁺CAR⁺ cells in their freshly prepared state.

CAR-T cells can sustain long-term anti-tumor effects in vivo through differentiation into memory subgroups [25]. A previous study has demonstrated that fresh PBMC predominantly adopt a central memory T cell (TCM) phenotype after undergoing cryopreservation [26]. Therefore, the expression of memory-associated and exhausted markers in both fresh and frozen states of both anti-HER2 and anti-HER2-13 CAR-T cells, derived from dual-donor provenance, was analyzed by flow cytometry. Comparative analysis failed to identify any significant disparities between the two CAR-T cell constructs containing different scFvs, derived from the same donor, in terms of effector memory T cell (TEM, CD62L^-CD44⁺) and central memory T cell (TCM, CD62L⁺CD44⁺) subpopulations, as well as in the expression of immunoexhaustion-associated molecules including PD-1, TIM-3 and LAG-3 (Fig. 3A,B).

In addition, the status of CAR-T cells, including cell viability, expression of exhaustion markers, and the percentages of CAR⁺CD3⁺, TCM, TEM phenotypes, were also evaluated after one-month or six-month cryopreservation. The data showed a decrease in the TEM subpopulation and an increase in the TCM subpopulation in both anti-HER2 and anti-HER2-13 CAR-T cells. At the same time, the expressions of exhaustion markers, such as LAG-3, TIM-3, and PD-1, in both CAR-T groups showed a tendency to increase with prolonged cryopreservation time (Fig. 3C,D).

As presented in Fig. 4A, SKBR-3 cells were identified as a type of breast cancer cell line with inherent HER2 overexpression. Nevertheless, HER2 protein expression was markedly reduced in SKBR-3 shHER2 cells relative to their wild-type counterparts. HER2 protein expression was significantly elevated in SPC-A-1^HER2+ and MDA-MB-231^HER2+ cells compared to their wild-type counterparts.

Compared with NT cells, both types of CAR-T cells co-cultured with magnetic beads effectively and specifically targeted HER2⁺ tumor cells, including SPC-A-1^HER2+, MDA-MB-231^HER2+ and SKBR-3 (Fig. 4B–D). Following 24 h of culture, both types of CAR-T cells significantly induced dose-dependent tumor cell death in all HER2 overexpression cancer cell lines. With an 8:1 ratio of CAR-T cells to target cells, anti-HER2 CAR-T cells exhibited over 80% cytotoxicity against all tested cancer cell lines, regardless of their HER2 expression status, indicating potential off-target effects. This finding indicated that anti-HER2 CAR-T cells possessed strong cytotoxic capabilities; however, their specificity toward HER2-positive tumor cells was limited. Meanwhile, anti-HER2-13 CAR-T cells demonstrated enhanced cytotoxicity toward HER2-positive cells compared with HER2-negative cells, especially at an effector-target ratio of 2:1 on SPC-A-1^HER2+/SPC-A-1 and MDA-MB-231^HER2+/MDA-MB-231 cells (p < 0.05, Fig. 4B,C). Nevertheless, both anti-HER2 and anti-HER2-13 CAR-T cells exhibited obviously higher tumoricidal effects on SKBR-3 cells compared to SKBR-3 shHER2 cells using a 2:1 or 4:1 effector-to-target cell ratios (p < 0.0001, Fig. 4D), likely due to the inherently high and widespread expression of HER2 in the former (Fig. 4A). Despite exhibiting weaker cytotoxicity (~60%) against SPC-A-1^HER2+ and MDA-MB-231^HER2+ cells compared to anti-HER2 CAR-T cells (~80%) at an E:T ratio of 8:1, anti-HER2-13 CAR-T cells may offer superior safety due to their greater specificity for HER2-positive targets.

To further investigate the anti-tumor efficacy of CAR-T cells and understand the underlying mechanisms, we performed ELISA experiments to measure the release of key cytokines, including IL-2, IL-6, IL-10, TNF-α, GM-CSF, perforin, and granzyme B from the supernatant of CAR-T cells and NT cells co-cultured with tumor cells at an E:T ratio of 2:1 for 24 h. Our results demonstrated that the differences in cytokine secretion among the two CAR-T cell types and NT cells co-cultured with tumor cells were in line with their respective cytotoxic effects against tumor cells. Specifically, both anti-HER2 and anti-HER2-13 CAR-T cells exhibited significantly elevated secretion of IL-2, IL-6, TNF-α, IFN-γ, and granzyme B relative to NT cells under 2:1 effector-to-target conditions (Fig. 5A,B,D,E,G). These data collectively imply that these cytokines play a crucial role in mediating the tumoricidal efficacy of CAR-T cells. Furthermore, both CAR-T groups also produced higher levels of IL-10, perforin, and GM-CSF by contrast to NT cells (Fig. 5C,F,H). The increased production of these factors by CAR-T cells indicated their possible involvement in immune regulation and enhancing overall tumor elimination.

Based on the promising results observed in in vitro studies, additional in vivo investigations were performed employing a xenograft tumor model (Fig. 6A). Our result demonstrated that the Trastuzumab group had a certain therapeutic effect relative to the control group, although it was not statistically significant (p = 0.0515, Fig. 6D). This result confirmed the successful establishment of our animal model, enabling the evaluation of different treatment modalities. Given that previous studies have demonstrated the beneficial effects of IL-2 on CAR-T cells or T cells [27,28], IL-2 was co-administered with CAR-T cells in the murine model. The data demonstrated that both CAR-T cell groups with IL-2 injection exhibited anti-tumor effects compared to the NT+IL-2 group, with the anti-HER2-13 CAR-T group showing a statistically significantly greater reduction in tumor weight (p = 0.0322) (Fig. 6C,D). Moreover, when examining the overall changes in body weight among the different treatment groups, we observed a temporary decrease in murine body weight in the anti-HER2 CAR-T treatment group. However, no statistically significant alterations in body weight were observed in the other treatment groups throughout the entire treatment period (Fig. 6B).

At the study endpoint, Ki-67 and TUNEL immunofluorescence staining were used to detect proliferating and apoptotic cells, respectively, in tumor tissue. The findings revealed a statistically significant rise in the frequency of apoptotic cells in the Trastuzumab group in comparison with the control group (Fig. 7C, p < 0.05). Furthermore, both CAR-T cell treatment groups exhibited a significantly higher number of apoptotic cells as opposed to the NT group (Fig. 7A,C), indicating that CAR-T cells could effectively induce tumor cell apoptosis and thereby exert their anti-tumor effects. The results of the Ki-67 assay showed that no significant differences in Ki-67 expression in tumor tissues among the Trastuzumab-treated group, the NT cell-treated group, and the control group. However, both CAR-T cell treatment groups exhibited statistically significant reductions in Ki-67 expression compared to the NT cells treatment group. This suggested that both CAR-T cells effectively suppressed the proliferative capacity of tumor cells (p < 0.05, p < 0.01), with the anti-HER2-13 CAR-T treatment group exhibiting a notably stronger effect (Fig. 7B,D).