HEK 293T packaging cells were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA); the triple-negative human breast cancer cell line MDA-MB-468 (abbreviated MDA) and its variant stably expressing HER2 (MDA-HER2) (11) were a kind gift of Stephen Gottschalk (St. Jude Children’s Hospital; Memphis; TN; USA). Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 2 mmol/l GlutaMAX, 10% Fetal Calf Serum (FCS) and antibiotics. Primary human T cells, UniCAR and HER2-specific CAR T cells were cultured in RPMI (Roswell Park Memorial Institute) medium supplemented with 2 mmol/l GlutaMAX, 10% FCS and antibiotics.

All cells and cell lines listed above were maintained in a humidified atmosphere containing 5% CO₂ at 37°C and were routinely checked for the absence of Mycoplasma contamination. MDA.ffLUC and MDA-HER2.ffLUC cell lines were generated by single-cell cloning of MDA-MB-468 and MDA-HER2 cell lines, respectively, after transduction with a retrovirus encoding eGFP.ffLUC to express an enhanced green fluorescent protein/firefly luciferase fusion gene (9).

Retroviral particles were generated by transient transfection of HEK 293T cells with CAR coding retroviral transfer vectors, the Peg-Pam-e plasmid containing the sequence for MoMLV gag-pol, and the pMax.RD114 plasmid containing the sequence for RD114 by using jetPrime transfection reagent (Poly-plus, Illkirch, France). The helper plasmids were kind gifts of Stephen Gottschalk (St.Jude Children’s Hospital; Memphis; TN; USA). To generate a pSFG retroviral UniCAR construct, we incorporated the mSA2-derived recognition domain (kindly provided by Jason J. Lohmueller (7)) into the CAR backbone consisting of the IgG heavy chain signal peptide; the IgG1 short hinge, a transmembrane region of human CD28 with CD28 intracellular costimulatory endo domain and the cytoplasmic region of human CD3ζ (12). The trastuzumab-derived HER2-specific CAR, a kind gift from Dr. Richard A. Morgan at the NCI, NIH, is coded in an MSGV vector and is composed of the Igκ chain signal peptide; trastuzumab light chain variable region; the 218 peptide linker; trastuzumab heavy chain variable region; the hinge, transmembrane and cytoplasmic regions of human CD28; and the cytoplasmic region of human CD3ζ (13, 14). Supernatants containing the retrovirus particles were collected after 48 hours.

Experiments were carried out on human samples in accordance with the Declaration of Helsinki and approved by the Regional and Institutional Committee for Research Ethics (RKEB.5378/2019). To generate CAR T cells, human peripheral blood mononuclear cells were isolated by Ficoll gradient centrifugation and stimulated in non-tissue culture 24-well plates pre-coated with 1 µg/ml OKT3 (Thermo Fischer, Waltham, MA, USA) and anti-CD28 (R&D Systems, Minneapolis, MN, USA) antibodies. On day 2, human interleukin-7 (IL-7; 10 ng/ml) and human interleukin-15 (IL-15; 5 ng/ml) (Miltenyi Biotec, Bergisch Gladbach, Germany) were added to cultures. T cells were transduced with retroviral particles on RetroNectin-coated (Takara, Kusatsu, Japan) plates on day 3 in the presence of IL-7 (10 ng/ml) and IL-15 (5 ng/ml). The expansion of T cells was subsequently supported with IL-7 and IL-15. OKT3/CD28 activated non-transduced (NT) T cells were expanded with IL-7 and IL-15 using the same protocol. Following 48h incubation, cells were used for further experiments.

In cell products, HER2-specific CAR expression was confirmed by labeling with a HER2-Fc fusion protein (R&D Systems, Minneapolis, MN, USA) followed by Alexa Fluor 647 conjugated anti-human IgG (Invitrogen/Thermo Fisher/, Carlsbad, CA, USA), UniCAR expression was confirmed by labeling with biotinylated highly cross-absorbed goat-anti-chicken IgG conjugated with Alexa Fluor 647. T cell purity was determined by Alexa Fluor 488 conjugated anti-human CD3 antibody (BD Biosciences, San Jose, CA, USA) staining.

In murine blood samples, human T cells were identified based on their pre-loaded CFSE signal and by labeling with APC-conjugated anti-human CD45 antibody (BD Biosciences, San Jose, CA, USA) antibody. The presence of conventional HER2-CARs or biotinylated trastuzumab-bound UniCAR molecules in the membrane of human T lymphocytes was confirmed by labeling with an Alexa Fluor 647 conjugated monomeric HER2 extracellular domain (Sino Biological Europe GmbH, Eschborn, Germany) in order to achieve comparable signals for the two cell types in a single labeling step.

Conjugation of the biotin and HER2 with Alexa Fluor 647 (Thermo Fisher Scientific, Waltham, MA, USA) was carried out according to the manufacturers’ specifications and as previously described (15). For the biotinylation of trastuzumab, a standard protein modification protocol was used. The antibody was dissolved in PBS (pH 7.0) at a concentration of 1 mg/ml. The molar ratio in the reaction mixture was 5:1 of biotinyliating reagent (NHS-Biotin 1 mg/ml in DMSO) to trastuzumab. The reaction took place under constant stirring at room temperature for 30 min. Biotinylated trastuzumab was purified using protein A affinity chromatography.

All molecules were used at 10 µg/ml final concentration for 10 minutes on ice. Analysis was performed on at least 10,000 cells per sample using a NovoCyte 3000RYB (ACEA Biosciences, San Diego, CA, USA) instrument and NovoExpress software (ACEA Biosciences, San Diego, CA, USA).

Expression of CARs was confirmed by western blot, using an anti-human CD3ζ antibody for detection. t. HER2-CAR, UniCAR and NT T cells were washed with PBS and whole cell lysates were prepared in a lysis buffer containing 50 mM Tris-HCl, 150 mM NaCl, 0.1% Triton X-100, 5 mM EDTA and a protease inhibitor cocktail of 2 mM PMSF, 1 mM sodium orthovanadate, and 1X cOmplete™ Mini Protease Inhibitor Cocktail tablet (Roche, Basel, Switzerland). Cell lysates were resolved on SDS-polyacrylamide gels in a reducing loading buffer containing 0.1 mM DTT and blotted to PVDF membranes. Membranes were blocked with 5% milk in Tris-buffered saline containing 0.1% Tween20 for 1 hour at room temperature and washed in PBS. The membranes were then probed with 1 µg/ml mouse anti-human CD3ζ (BD Biosciences, San Jose, CA, USA) antibody overnight at 4°C. Membranes were washed and probed with a secondary anti-mouse IgG–horseradish peroxidase (HRP) for 1 hour at room temperature.

HER2-specific CAR T cells or UniCAR T cells (in the presence or absence of various concentrations of biotinylated trastuzumab) were plated onto 1 µg/ml HER2-Fc (R&D Systems, Minneapolis, MN, USA) pre-coated plates or cocultured with MDA-HER2 target cells (in the presence or absence of 10 µg/ml biotinylated trastuzumab) at a 1:1 effector to target ratio. Following 24 hours of culture, the supernatant was harvested and analyzed for the presence of interferon-gamma (IFNγ) by ELISA (R&D systems, Minneapolis, MN, USA) according to the manufacturer’s instruction using a Spark® multimode microplate reader (Tecan Group Ltd., Männedorf, Switzerland). The HER2^- MDA-MB-468 cell line and NT T cells served as controls.

Cytotoxic activity of various CAR T cells against targets was determined by a luciferase-based cytotoxicity assay (9). MDA and MDA-HER2 cells expressing eGFP/ffLUC were plated in 96-well flat bottom plates at a concentration of 3 × 10⁴ cells/well in quadruplets. After 24h, various effector cells were added at a 1:1 effector to tumor cell ratio in the presence or absence of 10 µg/ml of biotinylated trastuzumab. Wells without effector cells served as untreated control references. After 24h, luciferase activity was determined using a luciferase assay kit according to the manufacturer’s instructions (Promega, Madison, WI, USA) and a Synergy HT luminometer (BioTek, Winooski, VE, USA).

HER2-specific CAR T cells or UniCAR T cells (in the presence or absence of 10 µg/ml biotinylated trastuzumab; 2 × 10⁵ cells/well) were placed onto flat-bottom plates pre-coated with 1 µg/ml HER2-Fc (R&D Systems, Minneapolis, MN, USA). Every 3.5 days, the effector cell number was determined by flow cytometry, and then the proliferation rate was calculated by dividing the total effector cell number on the present day by the number of effector cells plated at the beginning of the last 3.5-day round. A new round was initiated by re-plating 2 × 10⁵ effector cells onto freshly coated plates. The effector cells did not receive interleukin supplements during the experiment. If the proliferation rate fell below 1, then the available maximum amount of effector cells was placed onto fresh plates at the beginning of a new round. The experiment was concluded for any subset when the proliferation rate of the effector cells fell under 0.45.

MDA-HER2.ffLUC cells (1 × 10⁵ cells/ml in 200 μl) were placed into 96-well U-bottom plates in a cold medium containing 2.5% Matrigel (BD Biosciences, San Jose, CA, USA). The suspensions were centrifuged at 1000×g for 10 minutes at 10°C, and the cell pellet was cultured for 10 days. Cytotoxic activity of UniCAR T cells in these three-dimensional cell cultures was determined by propidium iodide incorporation assay (16). Spheroids of equal size were cocultured with 2 × 10⁵ effector cells. After 24 h, 3D cocultures were labeled with 1 µg/ml propidium iodide and target cell killing was measured with a Zeiss LSM 880 confocal microscope. A quantitative digital image processing pipeline (16), created in ImageJ (17), was used to calculate PI incorporation of MDA-HER2 cells in the entire spheroid and the inner core, which was defined to be within half the radius of the spheroid. PI incorporation was determined in these regions as integrated PI fluorescence intensity above the threshold in all background-corrected images. Two spheroids and at least five 2-µm-thick optical slices per sample were analyzed.

NSG mice received a single i.v. dose of 20 × 10⁶ NT T cells (NT group) or HER2-CAR T cells (HER2-CAR group), or UniCAR T cells (UniCAR group). Mice co-treated with UniCAR T cells and biotinylated trastuzumab were treated with 100 µg biotinylated trastuzumab in 100 µl PBS i.p. 15 hours prior to injecting a single i.v. dose of 20 × 10⁶ UniCAR T cells. All administered human T cell products were pre-labeled with 1µM CFSE for 10 minutes on ice.

Retroorbital blood samples were taken from each mouse 2 minutes after T cell injection, and the animals were sacrificed.

The blood samples were mixed with RBC lysis puffer, washed with PBS twice, separated into two equal samples, and stained for human CD45 and for CAR T cells. Samples were stained and analyzed by flow cytometry as described above.

At termination, mice were dissected, lung samples were embedded in cryomatrix (Thermo Fischer Scientific, Waltham, MA, USA), and snap-frozen in isopentane cooled with liquid nitrogen.

NSG (NOD.Cg-Prkdcscid/Il2rgtm1Wjl/SzJ) mice were purchased from The Jackson Laboratory and housed in a specific pathogen-free environment. All animal experiments were performed in accordance with FELASA guidelines and recommendations and DIN EN ISO 9001 standards. On day 0, each seven-week-old female NSG mouse participating in the study was given a subcutaneous injection in both flanks, each containing 3 × 10⁶ MDA-HER2.ffLUC cells in 100 µl PBS mixed with an equal volume of Matrigel (BD Biosciences, San Jose, CA, USA). Tumor growth was monitored with an IVIS Spectrum CT instrument (Perkin Elmer, Waltham, MA, USA). Before measurement, isoflurane-anesthetized animals were injected i.p. with D-luciferin (150 mg/kg). A bioluminescence image was obtained and analyzed after 10 minutes using Living Image software Version 4.0 (Caliper Life Sciences, Waltham, MA, USA). Signal intensity measured as total photons per second per square centimeter per steradian (p/(s x cm2 x sr)) was obtained from identically sized ROIs. Effector cell treated mice received on day 21 a single i.v. dose of 2 × 10⁶ NT T cells (NT group) or HER2-CAR T cells (HER2-CAR group), or UniCAR T cells (UniCAR group). Mice co-treated with UniCAR T cells and biotinylated trastuzumab received on day 21 a single i.v. dose of 2 × 10⁶ UniCAR T cells plus treated with 100 µg biotinylated trastuzumab in 100 µl PBS i.p twice weekly from this day (UniCAR+BT group). BT treated mice received 100 µg biotinylated trastuzumab in 100 µl PBS i.p. twice weekly from day 21 post tumor cell inoculation (BT group). ( Supplementary Figure 1 ). Experiments were approved by the National Ethical Committee for Animal Research (# 5-1/2018/DEMÁB).

At termination, mice were dissected, and fresh tumors and organs (heart, lung, liver, kidney, spleen) were embedded in cryomatrix (Thermo Fischer Scientific, Waltham, MA, USA) and snap-frozen in isopentane cooled with liquid nitrogen. Serial 4 µm thick cryosections were made with a Shandon Cryotome (Thermo Fischer Scientific, Waltham, MA, USA) at −24°C, and air-dried and stored at -20°C until further use. Cryosections were stained for human CD4⁺, CD8⁺ and CD45⁺ T cells.

All labeling molecules were diluted in PBS buffer supplemented with 1% BSA. After 5 min of rehydration of the cryosections in PBS buffer containing 1% BSA and 0.01% TritonX-100 (Thermo Fischer Scientific, Waltham, MA, USA), human T cells were stained with Alexa Fluor 647 conjugated rat anti-human CD8 (YTC 182.20; prepared from hybridoma supernatant; (18)), Alexa Fluor 488 conjugated rat anti-human CD4 (YNB 46.1.8; prepared from hybridoma supernatant; (18)) and APC conjugated anti-human CD45 (BioLegend, San Diego, California) antibodies. The possible presence of mouse T cells was investigated by staining with Alexa Fluor 647 conjugated rat anti-mouse CD8 (YTS 105.18.10; prepared from hybridoma supernatant) (19); Alexa Fluor 546 conjugated rat anti-mouse CD3 (145-2C11; prepared from CRL-1975 (ATCC) hybridoma supernatant) and Alexa Fluor 488 conjugated anti-mouse CD4 (YTS 177.96.1; prepared from hybridoma supernatant) (19). Mouse macrophages were stained with APC-conjugated F4/80 antibodies (BD Biosciences, San Jose, CA, USA), while mouse neutrophil granulocytes were stained with APC-conjugated Ly-6G rat antibodies (1A8-Ly6g, Thermo Fischer Scientific). HER2 expression was investigated with Alexa Fluor 647 conjugated anti-human HER2 mAb (ErbB2-76.5; produced from hybridoma supernatants (ErbB2-76.5, a kind gift from Y. Yarden, Weizmann Institute of Science, Rehovot, Israel)). The presence of biotin in organ samples was confirmed by FITC-conjugated extravidin. All antibodies were used at 10 µg/ml concentration on ice for 1 hour. Sections were washed for 5, 10, 5, and 10 minutes using DAPI at 1 µg/ml in the third wash buffer for nuclear staining and mounted in Mowiol antifade.

Samples were stained with the standard hematoxylin-eosin procedure and fixed with DPX. The prepared samples were scanned in with a Pannoramic confocal microscope at ×20 magnification. The composite images were handled by CaseViewer software (3DHISTECH, Budapest, Hungary).

UniCAR and HER2-CAR cocultures were analyzed with a Zeiss Axiovert 200M microscope (×10 objective). Images were taken with ZEN 2.6 Blue Edition software, postprocessing was handled with Fiji ImageJ 1.53t.

Fluorescence-labeled tissue sections were analyzed with a confocal laser scanning microscope (LSM 880, Carl Zeiss GmbH, Jena, Germany). Alexa Fluor 488 and EGFP were excited at 488 nm, Alexa Fluor 546 and propidium iodide (PI) at 543nm, APC and Alexa Fluor 647 at 633 nm. Corresponding fluorescence emission was separated with an appropriate quad-band dichroic mirror and detected on a 32-element GaAsP spectral detector in bands of 505 to 550 nm, 560 to 615 nm, and above 650 nm, respectively. For tissue samples, 5 consecutive, 2 µm thick optical sections were taken at 3 µm intervals, covering the central 10 µm part of the sections.

GraphPad Prism 5 software (GraphPad software, Inc., La Jolla, CA) was used for statistical analysis. Data were presented as mean ± SD or SEM. For comparison between two groups, a two-tailed t-test was used. One-way ANOVA with Bonferroni’s post hoc test was used to compare three or more groups. Survival, measured from the time of tumor cell injection, was analyzed by the Kaplan-Meier method and log-rank test. P values < 0.05 were considered statistically significant.

First, we aimed to investigate whether CARs with universal recognition capability, upon binding HER2 through their specific linker, could induce an anti-tumor response similar to conventional HER2-specific CARs. The trastuzumab-derived recognition domain (9, 12) is replaced by the affinity-enhanced monomeric streptavidin 2 (mSA2) biotin-binding domain with its optimized hinge (7), thus we have ensured that the difference between the evolved anti-tumor effects depends solely on the recognition domains. Thus, we generated a UniCAR construct (mSA2.CD28.z) that consists of the same CD28 transmembrane, CD28 costimulatory and CD3ζ signal domain as the conventional HER2-specific CAR (4D5.CD28.z) ( Figure 1A ). Both vectors were packaged into RD114 pseudotyped retroviruses and transduced into primary human T cells.

Mean transduction efficiency in the CD3⁺ human lymphocyte population was 73.4% in the UniCAR and 84.0% in the HER2-CAR group, as judged by flow cytometry on day 4 post-transduction ( Figures 1B, C ). Immunoblot analysis confirmed the presence of UniCAR and HER2-CAR derived CD3ζ subunits in transduced T cells ( Supplementary Figure 2 ).

In vitro activation, anti-tumor efficacy and long-term persistence of UniCAR T cells targeted by soluble biotinylated trastuzumab (BT) and HER2-CAR T cells targeted by a trastuzumab derived ScFv were investigated in activation, coculture and rechallenge experiments. First, in a cell-free activation assay, we confirmed that various concentrations of biotinylated trastuzumab could induce UniCAR T cell activation on plates coated with immobilized recombinant HER2-Fc chimera protein. From the amount of IFNγ secreted during 24h incubation, we concluded that in the presence of the HER2-specific soluble linker, UniCAR T cell activation was strictly antigen-dependent and correlated with the concentration of the linker. Without biotinylated trastuzumab, there was no cytokine secretion, confirming specificity ( Figure 1D , green area). However, conventional HER2-CAR T cells at the same target concentration secreted higher levels of IFNγ than UniCAR T cells with the highest (10 µg/ml) BT concentration ( Figure 1D , HER2-CAR). This maximal concentration of biotinylated trastuzumab could only induce limited UniCAR T cell activation in the absence of the HER2 target antigen, suggesting that soluble linkers alone have a weak capacity to organize UniCARs into supramolecular clusters (CAR synapse) and thereby initiate effector cell activation ( Figure 1D , blue area). We confirmed that this background activation is due to mSA2 crosslinking by BT carrying more than one biotin per antibody molecule (data not presented). Non-transduced T cells (NT) were not activated even in the presence of high doses of the BT and the HER2 target ( Figure 1D , NT).

In contrast to their subpar activity on panned molecular targets, in cocultures with trastuzumab-resistant HER2⁺ MDA-HER2 cells at a 1:1 effector to target ratio, biotinylated trastuzumab targeted UniCAR T cells secreted IFNγ at amounts comparable to conventional HER2-CAR T cells ( Figure 1E ; UniCAR+BT or HER2-CAR vs. all other samples: p < 0.001; UniCAR+BT vs. HER2-CAR: n.s.) and also induced killing of monolayer tumor cell cultures with similar potency ( Figure 1F ; UniCAR+BT or HER2-CAR vs. all other samples: p < 0.01; UniCAR+BT vs. HER2-CAR: n.s. and Supplementary Figure 3 ). There was no cytokine secretion or tumor lysis in the absence of target ( Figures 1E, F , MDA), CAR ( Figures 1E, F , NT) or BT ( Figures 1E, F , UniCAR).

Having explored the early activation and anti-tumor efficacy of UniCAR and conventional HER2-CAR T cells, we examined whether the indirect or direct HER2 binding impacts the long-term in vitro persistence of the CAR T cells. UniCAR T cells in the presence of 10 µg/ml biotinylated trastuzumab and, for comparison, HER2-CAR T cells were restimulated twice weekly on immobilized HER2-Fc molecules. NT T cells with or without BT and UniCAR T cells without BT served as controls. We found that the proliferative capacity of BT-targeted UniCAR T cells was comparable, although slightly lower than that of conventional HER2-CAR Ts ( Figure 1G ).

We have previously demonstrated that the MUC4 and CD44-rich extracellular matrix (ECM) in trastuzumab-resistant tumors, as in MDA-HER2, develops an impenetrable barrier to antibodies and thus effectively inhibits trastuzumab-dependent NK cell-mediated cytotoxicity. At the same time, conventional HER2-CAR T cells, as actively moving cell therapeutics, recognize the ECM-masked HER2 target antigen in the spheroid core and induce a potent cytotoxic response (9). The finding that BT-targeted UniCAR T cells recognized and killed MDA-HER2 in monolayer cultures encouraged us to investigate their anti-tumor functions in tumor spheroids with an established ECM. To explore whether UniCAR T cells can access ECM-masked HER2⁺ cells through a soluble HER2-targeting linker, we set up a coculture experiment in which MDA-HER2 spheroids were cultured with effector UniCAR T cells for 24h in the presence or absence of biotinylated trastuzumab. Using confocal microscopy (16), we visualized ( Figure 2A ) and analyzed ( Figures 2B, C ) tumor cell killing indicated by incorporation of propidium iodide (PI). UniCAR T cells in the presence of 10 µg/ml biotinylated trastuzumab showed substantially higher cytolytic activity in the inner spheroid core than in the absence of the BT linker (p < 0.001) ( Figure 2C ).

Given that UniCAR T cells alone do not penetrate spheroids ( Figure 2 ) and trastuzumab alone does not penetrate these spheroids (9), we hypothesized that for effective tumor penetration UniCAR T cells and BT must pre-assemble. To reveal whether this pre-assembly happens in the therapeutic setting, in our first in vivo experiment 20x10⁶ CFSE prelabeled UniCAR T cells were administered i.v. into an NSG mouse pretreated i.p. with 100 µg of biotinylated trastuzumab 15 h before effector cell injection. A blood sample was taken two minutes after tail vein injection, and the presence of biotinylated trastuzumab-bound UniCARs in the T cell membrane was determined using the Alexa Fluor 647-conjugated HER2 (A647-HER2). Conventional HER2-CAR T cells served as positive, NT and UniCAR T cells without BT pretreatment as negative controls. We found that almost immediately after UniCAR T cell injection, soluble BT bound to the mSA2 domain, as indicated by A647-HER2 binding to the CFSE⁺ population, similarly to the HER2-CAR positive control ( Figure 3A , UniCAR + BT: 33% vs. HER2-CAR: 21%). We confirmed that CFSE-positive cells are identical to the human CD45-positive cells ( Supplementary Figure 4 , upper panel, second row, CD45-APC vs. CFSE). However, we could recover considerably fewer human T cells from the UniCAR + BT mouse than from the other animals by sampling the same volume. Since upon CAR T cell injection, effector cells primarily gather in the lungs, with a notable presence in other organs beginning only 14 hours post-injection (20), in search of a possible explanation we have investigated the presence of injected cells in the lungs and found an significantly increased density of CD45-positive human T cells in the lungs of the UniCAR + BT mouse compared to the other animals ( Figures 3B, C , red columns). We observed a predominance of CD4+ T cells in the lungs of the UniCAR T + BT-treated mouse, whereas a balance of CD4+ and CD8+ T cells was observed in the lungs of HER2-CAR T cell injected animal. ( Figure 3C , yellow and green columns and Supplementary Figure 5 ). Interestingly, A647-HER2 binding was also observed in a CFSE^-/CD45^- population of UniCAR + BT treated mice, indicating the presence of mouse-derived, Fc-receptor expressing phagocytes that can bind trastuzumab ( Supplementary Figure 4 , bottom panel).

To compare the in vivo anti-tumor effect of UniCAR and conventional CAR T cells, subcutaneous HER2⁺ MDA-HER2.ffLUC xenografts were grown in NSG mice and treated with a single i.v. dose of 2 × 10⁶ UniCAR or HER2-CAR T cells on day 21 post tumor cell inoculation ( Figure 4A , Supplementary Figure 1 ). Except for the control group for off-target effects (UniCAR group), animals treated with UniCAR T cells also received 100 µg biotinylated trastuzumab in 100µl PBS i.p. on the day of CAR T cell injection and twice weekly thereafter (UniCAR+BT group). As the in vivo binding experiment ( Supplementary Figure 3 , bottom panel), our previous results (9) and the experience of others (21)have shown that NSG mice may have functional monocytes and macrophages, we investigated the potential antibody-dependent cellular cytotoxicity (ADCC) induced by biotinylated trastuzumab treatment in a linker only group (BT group). Mice treated with non-transduced T cells served as a reference group (NT group) ( Supplementary Figure 1 ).

Surprisingly, mice co-treated with UniCAR T cells and biotinylated trastuzumab started to die within a week of CAR T injection; moreover, all the animals were lost within four weeks, with no appreciable tumor regression. No similar effect was seen in any of the control groups. As expected from earlier experience, all the mice in the HER2-CAR T cell treated group were cured ( Figures 4A, B ).

Routine pathological H&E stained tissue sections of the treated animals revealed a massive cellular infiltration around the blood vessels in the lungs of mice co-treated with UniCAR T cells and biotinylated trastuzumab ( Figure 5A ; UniCAR+BT, yellow arrows). No abnormalities were found in the liver, spleen, kidneys, or heart ( Supplementary Figure 6 ).

Subsequently, a systematic immunohistochemical screen revealed biotin accumulating and trastuzumab binding cells in all treatment groups ( Supplementary Figure 7 ). In addition, in the lungs of UniCAR+BT treated mice we detected many CD4⁺ and CD8⁺ human lymphocytes on day 27 post-treatment, indicating the presence of UniCAR T cells ( Figure 5B , Supplementary Figure 7 ). We could also see human CD8⁺ lymphocytes in the lungs of NT, UniCAR and HER2-CAR T cell treated animals on day 64 post effector cell injection, indicating the emergence of xeno recognition ( Supplementary Figure 7 ; NT, UniCAR, HER2-CAR). However, in the lungs of UniCAR+BT-treated animals, the cellular infiltration and pathological effects appear to be mainly due to the presence of murine macrophages, which are detectable in high numbers along with human T cells only in these animals ( Supplementary Figure 8 ; UniCAR+BT). In contrast, in the lungs of HER2-CAR-treated mice, less intensive mouse macrophage infiltration was observed even at day 64 post effector cell injection, whereas the number of mouse neutrophil granulocytes was slightly increased ( Supplementary Figure 8 ; HER2-CAR). Less intensive macrophage or granulocyte infiltration was observed in the lungs of UniCAR-treated mice if no BT was present ( Supplementary Figure 8 ; UniCAR).

In tumors co-treated with UniCAR T cells and BT, we detected a high number of infiltrating human T lymphocytes as early as 27 days after inoculation ( Figure 6 , UniCAR+BT). We confirmed that the majority of tumor-infiltrating lymphocytes were CD4⁺ T cells. In contrast, in mice treated with UniCAR T cells only or with NT T cells only, on day 64 post effector cell injection tumors were mainly infiltrated with CD8⁺ T cells, which have possibly proliferated due to allorecognition of the human tumor tissue ( Figure 6 , UniCAR, NT). In the HER2-CAR-treated group, the rapid cytolytic effect did not allow tumor excision. Overall, these results demonstrate that UniCAR T cells coupled with biotinylated-trastuzumab can penetrate trastuzumab-resistant xenografts.