Leukapheresis units were obtained under written informed consent of the donors in accordance with study protocols approved by Hannover Medical School Ethics Review Board (approval no. 4837). All experiments involving mice were performed in accordance with the German Animal Welfare Act and were approved by the Lower Saxony Office for Consumer Protection and Food Safety (“Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit”, Dezernat 33/Tierschutz, LAVES; Protocol Nos. 33.12-42502-04-21/3791 and 33.12-42502-04-16/2347). Euthanasia was performed via cervical dislocation after CO₂ inhalation.

The HEK-293T cell line (abbreviated to 293T) was obtained from the American Tissue Culture Collection (CRL-11268™ ATCC, ATCC, Manassas, VA, USA). The lentivirally transduced 293T/gp350 cell line was previously established (15). 293T cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM; Gibco™ by Thermo Fisher Scientific, Waltham, MA, USA) with 10% heat-inactivated fetal bovine serum (FBS; Gibco™), 100 units/ml penicillin G, and 100 µg/ml streptomycin sulfate (1× P/S; Sigma-Aldrich, St. Louis, MO, USA). Nalm-6 (ACC-128 https://www.dsmz.de/collection/catalogue/details/culture/ACC-128 German Collection of Microorganisms and Cell Cultures (DSMZ), Brunswick, Germany), Jiyoye (ACC590, https://www.dsmz.de/collection/catalogue/details/culture/ACC-590, DSMZ), and Daudi (ACC78, https://www.dsmz.de/collection/catalogue/details/culture/ACC-78, DSMZ) lymphoma cell lines were cultured in Roswell Park Memorial Institute (RPMI) (Gibco™) with 10% FBS and 1× P/S. All cell cultures were maintained at 37°C and 5% CO₂. The cell lines were split every 2 to 3 days to maintain a density of 0.2–2.0 × 10⁶ cells/ml. Nalm-6, Daudi, and Jiyoye cell lines expressing fLuc-GFP were generated after transduction with the lentiviral vector pRRL.PPT.Cbx3-SFFV-fLuc-T2A-eGFP (the plasmid DNA containing the backbone vector was kindly provided by Prof. Axel Schambach, Institute of Experimental Hematology, Hannover Medical School, Germany). The cells were transduced with lentivirus for 24 h at 37°C and 5% CO₂. At 72 h after transduction, the expression of green fluorescent protein (GFP) was confirmed by flow cytometry using the BD LSR II (BD Biosciences, Franklin Lakes, NJ, USA). The GFP^bright population was sorted by the Cell Sorting Core Facility at the Hannover Medical School. GFP^bright Daudi and Jiyoye cells were expanded in RPMI with 10% FBS and 1× P/S at 37°C and 5% CO₂ and subsequently cryopreserved at −80°C.

Apheresis collection was performed at the Institute of Transfusion Medicine and Organ Engineering (Hannover Medical School). Peripheral blood mononuclear cells (PBMCs) of Donors 1 and 3 were purified by sediment centrifugation (Ficoll, Biochrome AG, Berlin, Germany) and cryopreserved, whereas the cells of Donor 2 were not purified prior to cryopreservation. PBMCs were cultured in RPMI with 10% FBS and 1× P/S and supplemented with interleukin (IL)-7 (5 ng/ml) and IL-15 (5 ng/ml) (both from Miltenyi Biotec, Bergisch Gladbach, Germany). Media and cytokines were exchanged every 2 to 3 days.

The CD19CAR HDRT sequence has been previously described (20) ( Figures 1A , S1A ). The CD19CAR incorporates a single-chain variable fragment (scFv) derived from the anti-CD19 FMC63 antibody, similar to currently approved CD19CAR-T cell products. FMC63 is a mouse IgG2a monoclonal antibody with an equilibrium dissociation constant (K _D) in the 10⁻⁴ nM range (22). The sequence of the CD19 HDRT is publicly available for academic and non-profit purposes (20) (see Addgene, https://www.addgene.org/183473/). The gp350CAR HDRT sequence was designed to incorporate the 7A1 scFv sequence targeting gp350. The parental 7A1 is a rat-derived monoclonal antibody, with high EBV in vitro neutralizing activity (15), but the K_D was not determined (Stripecke et al., patent pending; further information and materials are available upon execution of a material transfer agreement with the Hannover Medical School). We developed and tested an HDRT named gp350CAR-BE with the same backbone structure as CD19CAR (20). In addition, we created the HDRT gp350CAR-HA with the same gp350CAR backbone previously validated in retroviral vectors (15) ( Figures 1A , S1A ). The plasmids containing the gp350CAR HDRT sequences were synthesized by Twist Bioscience (South San Francisco, CA, USA). The HDRT templates were amplified from the plasmids by PCR using Herculase II Fusion DNA Polymerase essentially as described (20) and following the vendor’s instructions (Agilent, Santa Clara, CA, USA). The PCR was performed with 3 ng plasmid in 50-µl reactions using primers homologous to the extremities of the TRAC homology arms (forward 5′-ataaaagaataagcagtattattaagtagccctgc-3′; reverse 5′-atctgcttttttcccgtgtcattct-3′). For amplification, the samples were placed in a PCR cycler and heated to 95°C for 2 min for initial denaturation; this was followed by 30 cycles of 20 s at 95°C, 20 s at 59°C, and 90 s at 72°C. A final step of extension at 72°C for 3 min was performed, before storage at 4°C until purification with DNA-binding magnetic beads (AMPure XP, Beckman Coulter, Brea, CA, USA). The beads were pre-warmed to room temperature (RT) for approximately 30 min, suspended by vortexing, and mixed one-to-one with up to 600 µl PCR product in a 1.5-ml tube. After 10 min of incubation at RT, the tube was placed on a DynaMag™-2 magnet (Thermo Fisher Scientific) for an additional 10 min at RT. The supernatant was removed while the tube was still on the magnet; 1.2 ml of 70% ethanol was added, and the sample was mixed by inverting the entire magnet. After 3 min of incubation, the ethanol was removed, and the washing step was repeated once. After the second wash, the pellet was air-dried for approximately 10 min to a semi-dry state. The tube was removed from the magnet, and the pellet was resuspended in 3 µl nuclease-free water per 100 µl used PCR product. After 2 min of incubation at RT, the tube was incubated for 5 more minutes on the magnet. The supernatants were collected without touching the bead fraction and diluted to 1:20 dilution in water, and the concentrations of the HDRTs were measured using a Nanophotometer^® (Implen, Munich, Germany). The DNAs were then adjusted to a concentration of 1 µg/µl with nuclease-free water and stored at −20°C until use.

PBMCs were activated with TransAct (Miltenyi Biotec) for 50–54 h using 10 µl of TransAct per 2 × 10⁶ cells per ml RPMI with 10% FBS and 1× P/S, and supplemented with IL-7 and IL-15 (5 ng/ml each). Cells were washed once with PBS and resuspended at 1 × 10⁸ cells/ml in TexMACS (Miltenyi Biotec). We used 50-µl cuvettes for small scale (SS) production (5 × 10⁶ cells per electroporation) and 1-ml cuvettes for large scale (LS) production (1 × 10⁸ cells per electroporation). To generate the RNP complex, 3 µM of crRNA TRAC guide 6 AGAGTCTCTCAGCTGGTACA (100 µM; IDT, Newark, NJ, USA) was mixed with 3 µM of tracrRNA (100 µM; IDT, catalog no. 1072532); the RNAs were diluted in a nuclease-free duplex buffer provided by IDT and kept cryopreserved in aliquots at −20°C. The RNAs were incubated at 95°C for 5 min to form the guide RNA. After being cooled to RT, 2.5 µg/ml 15–50 kDa poly-l-glutamic acid (PGA) (Sigma-Aldrich; 0.8:1 ratio PGA:gRNA) and 1 µM Cas9 (Alt-R S.p Cas9 Nuclease V3, 61 µM, IDT) were added, and the mixture incubated for 30 min to allow RNP formation. Subsequently, 50 µg/ml HDRT was added to the RNPs. R-50x8 processing assemblies (ER050U8-03, MaxCyte, Rockville, MD, USA) were used for SS and R-1000 processing assemblies (ER001M1-10, MaxCyte) for LS electroporation. An ExPERT ATx^® device (MaxCyte) was used for electroporation with the “Resting T cells 14.3” program. The electroporated cells were transferred into either a 96-well plate (Sarstedt, Nümbrecht, Germany; SS) or a 24-well plate (Sarstedt; LS) for recovery. The cuvettes were washed with 50 µl (R-50x8) or 1 ml (R-1000) TexMACS medium (Miltenyi Biotec) containing 2 µM HDR Enhancer V2 (IDT) and added to the corresponding wells, resulting in a final enhancer concentration of 1 µM. After 30 min recovery at 37°C, the cells were transferred into a 24-well plate (SS) or a T-75 flask (LS; Nunc-Thermo Fisher Scientific, Kamstrupvej, Denmark) containing the appropriate volume of pre-warmed RPMI to dilute the cell density to 2 × 10⁶ cells/ml. After 2 h at 37°C, cytokines (IL-7 and IL-15, each 5 ng/ml) and 10% FBS were added, and the cells were further cultured as described previously for T cells (the next medium exchange contained 1× P/S). An interim analysis was performed on day 4 after electroporation to confirm the KO ( Figure S1 ). The time of analyses for KI was day 9 after electroporation, and for some experiments, cells were expanded for up to 14 or 19 days. Expanded cells were either freshly used or cryopreserved in freezing medium (20% FBS, 10% dimethyl sulfoxide (DMSO) in RPMI) and maintained at −150°C until being thawed for further use. For quality control, the gene-edited CAR-T cells were analyzed by flow cytometry for CD3-KO, CAR-KI, viability, and expression of CD4 and CD8 using an LSR II flow cytometer (BD). Cells were blocked in 100 µl PBS containing 10 µg/ml mouse IgG (Sigma-Aldrich). CAR KI was detected by staining with 1:50 Alexa Fluor^® 647 “AffiniPure F(ab’)₂ Fragment Goat Anti-Human IgG” (109-606-170, Jackson Immuno Research Laboratories, West Grove, PA, USA). The staining of the CAR with the anti-human IgG was performed separately from the other subsequently used staining monoclonal antibodies (mAbs) to avoid complex formation. Table S1 describes the antibodies used for staining. The cells were stained for 30 min on ice in darkness, washed twice in PBS containing 1% FBS, and resuspended in PBS. The data were acquired with an LSR II flow cytometer (BD Biosciences) and analyzed with FlowJo (Treestar).

The specificity and functionality of gp350^KICAR-T cells were tested through co-culture with 293T/WT or with 293T/gp350 cells. Target cells were seeded in a 96-well flat-bottom plate (1 × 10⁴ cells per well in 50 µl DMEM with 10% FBS and 1× P/S). The next day, we calculated the amount of effector ^KOTCR^KICAR-T cells added to the culture at 1:1 and 3:1 effector to target (E:T) ratios. All cultures were performed in triplicate and were maintained at 37°C and 5% CO₂ for 48 h. The supernatants of the triplicates were harvested and pooled, and IFN-γ secretion was analyzed using an ELISA kit in accordance with the manufacturer’s instructions (Human IFN-γ uncoated ELISA, Invitrogen, Carlsbad, CA, USA). Additionally, the concentrations of several human cytokines detectable in culture supernatants (GM-CSF, IFN-γ, IL-2, IL-5, IL-8, IL-10, IL-17A, IP-10, MCP-1, and TNF-α) were determined using a bead array Luminex kit in accordance with the manufacturer’s protocol (Milliplex Millipore, Burlington, MA, USA). The cells from the triplicate cultures were harvested, pooled, and washed with PBS, and the cell pellets were blocked in PBS containing 10 µg/ml mouse IgG (Sigma-Aldrich). To determine the number of viable cells, the cells were analyzed by flow cytometry (see Figure S2 ).

For luminometry and ELISA analyses, Daudi/fLuc-GFP and Jiyoye/fLuc-GFP were seeded in 96-well round-bottom plates (1 × 10⁴ cells per well). ^KOTCR^KICAR-T cell effectors were added to the culture at 1:1, 3:1, and 10:1 E:T ratios in 50 µl RPMI with 10% FBS and 1× P/S supplemented with IL-7 and IL-15 (both 5 ng/ml) per well. All co-cultures were performed in triplicate and were incubated for 72 h (at 37°C and 5% CO₂). After incubation, the cells were spun down. The cell supernatants for each triplicate were collected for IFN-γ analyses using an ELISA kit (Human IFN-γ uncoated ELISA, Invitrogen, Waltham, MA, USA) in accordance with the manufacturer’s instructions. The cell pellets were resuspended in 100 µl d-luciferin potassium salt solution (SYNCHEM, Felsberg, Germany) (2.5 mg in 1 ml PBS). Luminescence emitted by living cells was measured for each instance with a TriStar2 instrument (Berthold Technologies, Bad Wildbad, Germany). The functionality of the large-scale cryopreserved/thawed CAR-T cells was also analyzed by flow cytometry. For this purpose, 1 × 10⁶ lymphoma target cells were seeded in a 12-well plate, and CAR-T cells were added at a 1:1 E:T ratio (adjusted to the CD3⁻CAR⁺ population) in 500 µl RPMI with 10% FBS and 1× P/S supplemented with IL-7 and IL-15 (both 5 ng/ml) per well. The co-cultures were incubated for 72 h (at 37°C and 5% CO₂). The cells were harvested, washed with PBS, and spun down. The pellets were blocked in 100 µl PBS containing 10 µg/ml mouse IgG (Sigma-Aldrich). Cells were stained with 1:200 Fixable Viability Dye eFluor™ 450 (FVD450, 65-0863-14; Invitrogen) for viability control and with 1:400 PerCP anti-human CD4 (317432, BioLegend, San Diego, CA, USA) and 1:200 PE-Cy7 anti-human CD8 (300914, BioLegend). Viable cells (FVD450⁻), target cells (GFP⁺), and CD4⁺ or CD8⁺ CAR-T cells were identified by flow cytometry using the LSR II device (BD) (see Table S1A for antibodies and Figure S3 for gating strategy details).

First, 1 × 10⁶ CAR-T cells in 1 ml R10 were seeded in a 6-well plate, to which 1 × 10⁶ GFP⁺ target cells in 1 ml R10 were added. IL-7 and IL-15 were added at 5 ng/ml each, and the plates were incubated at 37°C and 5% CO₂. At each time point of the culturing process (24, 48, and 72 h), 0.4 ml of the cell suspension was taken for cell counting, staining, and flow cytometry analysis. For BM analyses, cryopreserved samples were thawed, maintained in culture for 2 h in R10, and analyzed. For staining of exhaustion markers, the cells were centrifuged at 300 g for 5 min and resuspended in PBS. Since anti-Fc CAR antibodies can potentially bind to other antibodies, the staining procedure was performed in two steps. First, the samples were blocked with mouse IgG (Sigma Aldrich) for 10 min on ice, and CAR staining was then performed for 30 min at RT. The cells were next washed with FACS buffer (PBS with 1% FCS) and centrifuged. Subsequently, the samples were resuspended in a master mix containing all other antibodies and stained for 30 min at RT (see Table S1B for antibodies and Figure S6 for gating strategy details). Finally, cells were washed with FACS buffer, centrifuged, and resuspended in PBS. For staining of regulatory T cells (T_regs), the samples were collected in FACS tubes, centrifuged at 300 g for 5 min, and resuspended in PBS. First, the samples were blocked with FcR (Miltenyi Biotec, cat. #130-059-901) for 10 min, and the cells were stained for 20 min at RT for detection of CD4. Intracellular labeling of FoxP3 was carried out using the IntraPrep Permeabilization Reagent kit (Beckman Coulter, cat. #A07803) in accordance with the manufacturer’s instructions (see Table S1C for antibodies and Figure S7 for gating strategy details). All samples were washed twice with PBS and measured using a MACSQuant X Flow Cytometer, and the analyses were performed using FlowJo software (version 10.8).

NOD.Cg-Rag1 ^tm1Mom IL-2Rγc ^tm1Wjl (NRG) mice were obtained from the Jackson Laboratory (JAX, Bar Harbor, ME, USA) and bred under pathogen-free conditions or used for experiments. Mice aged 6–12 weeks were used in the present study. In experimental group mice, 2 × 10⁵ Daudi/GFP-fLuc or 2 × 10⁵ Jiyoye/GFP-fLuc cells were injected intravenously (i.v.)into the tail vein; control mice were injected with PBS. Bioluminescence imaging (BLI) analyses were performed with the IVIS Spectrum CT apparatus (PerkinElmer, Waltham, MA, USA), and the data were analyzed using LivingImage software (PerkinElmer) as previously described (20). Briefly, mice were anesthetized using isoflurane and injected intraperitoneally (i.p.) with 2.5 mg d-luciferin potassium salt dissolved in 100 µl PBS. The mice were imaged in frontal and lateral view with an automated exposure time of up to 5 min. Female mice showed higher rates of engraftment and tumor outgrowth (see Figure S4 ) and were thus used for testing CAR-T cells. The first imaging analysis was performed at 4 days after lymphoma challenge (D4) to confirm tumor engraftment, to determine the baseline for each mouse, and for allocation of mice to experimental groups along with others with similar engraftments. Additional sequential analyses were performed at 2, 3, and 3–4 weeks after lymphoma challenge. After euthanasia, macroscopic analyses were performed with the open abdomen and chest, and imaging analyses were performed post-mortem for direct visualization of tumor dissemination in organs. Samples of peripheral blood (PBL), bone marrow (BM), spleen (SPL), and lymph nodes (LNs) were collected for flow cytometry for characterization of GFP, CD20, CD19, and gp350 expression (gating strategy details are shown in Figure S4 ).

Lymphoma engraftment was determined by BLI on D4. Mice were distributed into cohorts with comparable engraftments, and 1 × 10^{6 KO}TCR^KICAR-T cell effectors were injected i.v. During the experimental period of 3–4 weeks, the mice were monitored (depending on clinical presentation) daily or at least three times per week for weight, morbidity, and any clinical sign of graft-versus-host disease (GvHD). BLI analyses were performed weekly. The experiment was terminated if mice met the LAVES-approved euthanasia criteria (body weight reduction of ≥20%, ruffled fur, reduced activity, and/or self-isolation). After euthanasia, macroscopic examinations of the SPL, LNs, kidneys, and ovaries were performed, and the number of tumors was counted. The BM was flushed from both femora and pooled for each mouse. Some BM cells were snap-frozen for isolation of DNA, some were used for BM smears, and the rest were used for flow cytometry analysis or cryopreserved. The LNs and SPL were smashed through a 100-µm cell strainer to obtain single-cell suspensions. SPL and PBL were resuspended in erythrolysis buffer (0.83% ammonium chloride, 20 mM HEPES [pH 7.2]), incubated for 5 min at RT, and then washed with PBS. The number of viable cells in each tissue sample was quantified by Trypan blue exclusion. The cells were analyzed fresh or cryopreserved in freezing medium (20% FBS and 10% DMSO in RPMI) and maintained at −150°C.

Two panels were used for flow cytometry analyses: 1) to quantify the GFP⁺ lymphoma load and determine the expression of CD19 and gp350 antigens on the surface of GFP⁺CD45⁺CD20⁺ lymphoma cells (7A1 mAb + secondary Ab; or, to determine the background, secondary Ab only) and 2) to identify and quantify CAR-T cells (CD45⁺GFP⁻, CD4, CD8, and CAR). The list of antibodies used and amounts used per staining are shown in Table S1A , and gating strategies are shown in Figures S4, 5 . Approximately 5 × 10⁵ cells (lymphocytes isolated from mice, in vitro cultured tumor cell lines, or gene-edited T cells) were blocked for 30 min on ice in 100 µl of 10 µg/µl mouse IgG (Sigma-Aldrich) in PBS and then washed with 1% FBS in PBS. Analyses were performed using a BD LSR II flow cytometer.

The Giemsa staining method used for BM smears is described in the Supplementary Material . The morphology of bone marrow was assessed with an Olympus BX51 (Olympus, Tokyo, Japan) microscope and a 40×/0.75 numerical aperture objective or a 100×/1.3 numerical aperture objective with Zeiss Immersol Medium (Zeiss, Jena, Germany). OlympusXC50 (Olympus) and analySIS software (Soft Imaging System, Stuttgart, Germany) were used to capture images.

DNA isolation of mouse BM to determine the EBV load was performed with a QIAamp DNA Blood Mini Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s protocol. DNA concentration and purity were measured using the Nanodrop instrument. The DNA samples showed similar concentrations, and 10 μl was used per reaction. The EBV DNA detection by real-time PCR was performed with the EBV in vitro diagnostics PCR kit (GeneProof^®, Dolni Herspice, Brno, Czechia) after amplification of a specific conserved DNA sequence of the single-copy gene encoding the nuclear antigen 1 (EBNA1) and measurement of fluorescence emission. An internal standard (IS) was included in the reaction mix as a control for the PCR reaction. The detection kit explores the “hot start” technology, minimizing non-specific reactions and assuring maximum sensitivity. The analyses were performed using a LightCycler 2.0/480 instrument. The detection limit of the assay was approximately 1.9 × 10⁵ EBV units in the sample.

For data obtained in in vitro killing assays, t-tests were used to compare the experimental groups with controls. For in vivo data, the analyses were performed by a biostatistician (S. R. Talbot, co-author), and some datasets were analyzed using ANOVA or Kruskal–Wallis (in either case, with Tukey’s post-hoc testing to compare the three groups), or via a binomial regression, which was used for statistical analyses and group comparisons. Additional methods and references for statistical analysis are described in the Supplementary Material. The significance level was set to 0.05. Statistical analyses were carried out using GraphPad Prism V7.0 software (GraphPad Software, La Jolla, CA, USA) and R software V4.1.0 (R Foundation for Statistical Computing, Vienna, Austria).

In a previous study, Kath et al. presented technical improvements for fast and efficient virus-free insertion of CARs into the TRAC locus of T cells (20). The HDRT used for generation of CD19-specific gene-edited CARs contained a CD28 costimulatory domain, followed by the CD3 zeta effector chain ( Figure 1A ; see additional elements explained in legends). Here, we designed and tested two HDRTs for redirecting the CAR-T cells against the EBV/gp350 protein ( Figures 1A , S1A ). We compared a gp350CAR-BE HDRT containing the CD19CAR backbone with a gp350CAR-HA HDRT, analogous to a backbone previously validated in a retroviral vector (15). The gp350CAR-HA HDRT incorporating additional linkers and a different membrane export signal resulted in higher CAR expression ( Figure S1 ); therefore, it was used in further experiments. For the current set of studies, we adapted the cell processing steps ( Figure 1B ) in terms of upscaling capability with the ExPERT ATx^® system. This electroporation system is modular and can be developed from the preclinical small scale (SS, for in vitro testing) and preclinical large scale (LS, for in vitro and in vivo testing) toward an up-scaled clinical scale (for use in patients). The gene-edited cells were harvested after expansion, analyzed for identity, and tested in vitro or in vivo. SS productions with T cells obtained from three different donors were performed in small electroporation cuvettes (50 µl, 5 × 10⁶ cells; Figure 1C , see gating strategy used for flow cytometry analyses in Figures S1B, C ). LS productions were generated with T cells from a single donor (Donor 1) and were performed in larger electroporation cuvettes (1 ml, 1 × 10⁸ cells; Figure 1D ). For both the SS and LS methods, the knock-outs of the TCRα chain resulted in major reductions in CD3 detection on the cell surface ( Figures 1E, I ). Within the CD3⁻ population, CAR KI was variable within donors used for the SS and within CARs (~3%–5% for CD19^KICAR and ~10%–15% for gp350^KICAR; Figure 1F ). The PBMCs obtained from Donors 1 and 3, which were purified by Ficoll sediment centrifugation and used as a source of T cells, showed a better KI than the PBMCs from Donor 2, which were not purified and showed lower cell viability. KI efficiency was higher for the LS runs and comparable for both CARs (~20%, Figure 1J ). At day 9 of cell culture, T cell recovery was increased on average threefold relative to input for SS and LS ( Figures 1G, K ), resulting in total yields of ~1.5 × 10⁷ cells for SS ( Figure 1H ) and ~3.5 × 10⁸ total cells for LS ( Figure 1L ). After these key adaptation steps, the functional properties of ^KICAR-T cells were analyzed.

As a first in vitro analysis of the cytotoxic potency and specificity of gp350^KICAR-T cells, we used a 293T cell line with stable expression of gp350 (15) ( Figure 2A ). Co-cultures of SC CD19^KICAR-T or gp350^KICAR-T with 293T/WT or 293T/gp350 as target cells were performed with E:T ratios of 1:1 and 3:1 for 48 h ( Figure 2B ). The co-cultures were then stained for detection of CD45 (a pan-hematopoietic marker not expressed on 293T cells) and 7AAD to determine the fractions of dead and viable cells (see gating strategy details in Figure S2 ). Co-culturing of CAR-T cells against either CD19 or against gp350 with 293T/WT cells resulted in a proportion of dead target cells of approximately 20%, which was comparable with the proportion of dead cells in the culture of target cells without CAR-T cells ( Figure 2C ). Co-culture of 293T/gp350 with gp350^KICAR-T cells showed 40%–60% killing, whereas the other co-cultures showed approximately 20% background loss of survival. Supernatants were harvested, and the secreted cytokines were quantified by ELISA or bead array. Only the co-culture of gp350^KICAR-T cells with 293T/gp350 showed elevated levels of IFN-γ, GM-CSF, IL-2, IL-5, IL-8, IL-10, IL-17A, MCP-1, and TNF-α secretion ( Figures 2D, E ). These results confirmed the specificity of gp350^KICAR-T cells and indicated an E:T dose-dependent effect for killing upon exposure to gp350⁺ cell targets.

As the next step, we evaluated the capacity of CD19^KICAR-T and gp350^KICAR-T cells to react in vitro against BL cells infected with EBV. Humans can be infected with two types of EBV: type 1 (T1) and type 2 (T2). T1 EBV transforms B cells more efficiently than T2 EBV in vitro, and T2 EBV-infected B cells can produce more lytic virus. As putative cell line models derived from pediatric BL patients, we obtained commercially CD19⁺CD20⁺EBV⁺ human B-lymphoblastoid T1 Daudi (23) and T2 Jiyoye cells (24). Both cell lines are considered prototypic mature B cells and, as such, express costimulatory ligands such as CD80 and CD86. The Daudi cell line is positive for EBV/EBNA by PCR analyses and expresses the mRNA for the proto-oncogene BCL-2, but the BLZF-1 immediate-early protein and capsid protein had previously not been detectable by Western blotting after phorbol ester/sodium butyrate stimulation. Although EBV infection in Daudi was classified as latent, Daudi cells express glucocorticoid receptors (GRs) and show a glucocorticoid dose-dependent upregulation of BZLF1 mRNA expression by hydrocortisone (HC) and dexamethasone (Dex). Dex induces expression of the early gene products BLLF3 (encoding for the EBV dUTPase) and BALF5 (encoding for the EBV DNA polymerase) and thus can promote EBV reactivations (25). The Jiyoye cell line is positive for EBV/EBNA by PCR analyses, carries the t(8:14) translocation rearranging IGH with MYC, and expresses the BLZF-1 immediate-early protein after phorbol ester/sodium butyrate stimulation and Western blot analyses. The EBV infection in Jiyoye was classified as potentially lytic with the possibility of production of active virus. In order to set up in vitro and in vivo assays, Daudi and Jiyoye cell lines co-expressing fLuc and GFP were generated using lentiviral transduction. Co-expression of CD19 and the reporter GFP was confirmed by flow cytometry ( Figure 3A , left panels). When gated on the GFP⁺ cells, approximately 45% of the Daudi/fLuc-GFP cells and 10% of the Jiyoye/fLuc-GFP cells expressed gp350 ( Figure 3A , right panels). Within the bright and homogeneously expressing CD19⁺ population (corresponding to >95% of the BL cells), 61% of the Daudi/fLuc-GFP and 29% of the Jiyoye/fLuc-GFP expressed gp350, but at much lower and variable levels ( Figure 3B ). It is known that EBV reactivation can be induced in BL cells, for example by treatment with phorbol esters (e.g., phorbol myristate acetate [PMA]). One explanation for the variable and relatively low levels of gp350 expression is that cells displaying very high lytic reactivation might be more prone to modulations of cell cycle control, although the specific functions of gp350 as an apoptotic or anti-apoptotic factor remain to be fully determined (26). Both target cell lines were cultured for 3 days with CD19^KICAR-T cells or gp350^KICAR-T cells in different E:T ratios (1:1, 3:1, and 10:1) ( Figure 3C ). The in vitro assay to test the CAR-T cells utilized fLuc as a vitality reporter gene followed by luminometry, as viable cells produce high levels of bioluminescence while dead cells do not ( Figure 3C ). In addition, T cell activation was followed by measurement of IFN-γ secreted in the supernatant ( Figure 3C ). CAR-T cells produced as SS from Donors 1 and 3 were used for co-culturing in order to evaluate donor-dependent differences. CD19^KICAR-T cells produced from both donors greatly reduced the number of living Daudi/fLuc-GFP and Jiyoye/fLuc-GFP cells after 3 days of co-culture at all tested E:T ratios (1:1, 3:1, and 10:1) ( Figures 3D, F ). This was associated with a dose dependent secretion of INF-γ detectable in supernatants ( Figures 3H, J ). The co-culture gp350^KICAR-T cells with Daudi/fLuc-GFP or Jiyoye/fLuc-GFP target cells stimulated modest but dose-dependent cytotoxic effects ( Figures 3E, G ) and detectable but low levels of INF-γ ( Figures 3I, K ). In sum, SS CD19^KICAR-T cells showed robust cytotoxic effects against both types of lymphomas, while the functional effects of gp350^KICAR-T cells were low.

One important aspect in banking of ^KICAR-T cells is that they must remain unaffected by cryopreservation and thawing. Therefore, we further evaluated the properties and functionality of cryopreserved/thawed ^KICAR-T cells produced at LS from two different donors. CAR expression was observed in the range of 13%–20% after thawing ( Figure 4A ). The functionality was determined by two analytical assays: one using luminometry and the other based on flow cytometry ( Figure 4B , see gating strategy details in Figure S3 ). The luminometry-based assays performed with Daudi/fLuc-GFP or with Jiyoye/fLuc-GFP target cells showed significant dose-dependent killing results for CD19^KICAR-T ( Figures 4C, E ) and gp350^KICAR-T ( Figures 4D, F ). CD19^KICAR-T cells were more efficient in killing than gp350^KICAR-T cells. The flow cytometry-based assay confirmed efficient killing for both types of ^KICAR-T cells, with high viability of the effector cells ( Figures 4G–J ). Furthermore, the effector cells remaining in the cultures could be distinguished by flow cytometry as CD4⁺ (T helper type, T_h) and CD8⁺ (cytotoxic T lymphocyte (CTL) type; see gating strategy details in Figure S3A ). For all in vitro cultures, higher frequencies of CTLs were observed ( Figures 4K–N ). In sum, these results showed better killing performance of large-scale ^KICART-T cells and dominance of CTLs in vitro.

In the next step, we established BL xenograft models in order to evaluate the therapeutic effects of ^KICAR-T cells in vivo. The Daudi/fLuc-GFP model was established first. After pilot experiments to set up the lymphoma challenge, we selected a challenge format consisting of 2 × 10⁵ tumor dose i.v., as the full course of lymphoma development in mice could be followed by BLI from day 4 after challenge for up to 4 weeks ( Figure 5A ). The terminal analyses were then set at 4 weeks after challenge, as at this point the challenged mice started to meet the predefined euthanasia criteria in the approved animal protocols (see Material and Methods section). BLI analyses on day 4 post-challenge showed signals in the anatomical regions of the bones and liver ( Figure 5B ). Two weeks after challenge, the BLI signals were distributed throughout the body of the mice. After 4 weeks, the signals were highly perceptible in several anatomical regions, and particularly in the femurs, hips, and ribs. Female mice showed higher signals than male mice ( Figure 5B ). After euthanasia, BLI was repeated for direct visualization of the abdominal cavity. Macroscopic examination combined with BLI demonstrated that the signals were high in the LNs, kidneys, and ovaries ( Figure 5C ). Tumors could be sporadically visualized in the kidneys ( Figure 5D ). Age-matched non-challenged control mice showed no tumor development. BM smears stained with Giemsa showed the presence of small lymphocytes in control NRG mice and of large blasts in mice challenged with Daudi ( Figure 5E ). Flow cytometry analyses of SPL and BM samples confirmed the detection of GFP⁺ cells. LNs were easily detectable and largely infiltrated with GFP⁺ blasts ( Figure 5F ). After gating on the GFP⁺ cells, flow cytometry analyses further showed the persistent expression of the B-cell markers CD19 and CD20 and, to a lesser extent, of gp350 on the cell surface of the in vivo grown lymphoma ( Figures 5G, H , see gating strategy details in Figure S4 ). Next, the Daudi/fLuc-GFP model was used to evaluate the efficacy of ^KICAR-T cells in vivo ( Figure 6A ). LS ^KICART-T cells were freshly produced, and expression of CAR, CD8, and CD4 was determined during the cell culture process ( Figure 6B ). In parallel, female mice were challenged with 2 × 10⁵ lymphoma cells. After the engraftment was confirmed on day 4, the cohorts were set up based on the BLI signals. Each mouse was infused with 1 × 10^{6 KO}TCR^KICAR⁺ T cells. Mice were longitudinally monitored by BLI analyses and by weight, and were sacrificed at week 4 after challenge ( Figure 6A ). All challenged mice developed lymphomas, and the cohort treated with CD19^KICAR-T cells showed significant delays in BLI signal distribution compared with non-treated mice or mice treated with gp350^KICAR-T ( Figures 6C, D ). All the mice treated with CD19^KICAR-T cells showed partial responses (PR) ( Figures 6C, D ), and mice maintained constant weight gain overall. In contrast, the non-treated mice and mice treated with gp350^KICAR-T cells showed significant weight loss, indicating morbidities ( Figure 6E ). Flow cytometry analyses to detect the presence of GFP⁺ blasts in PBL, BM, and SPL confirmed the therapeutic effects promoted by CD19^KICAR-T cells ( Figure 6F ). Analyses of EBV DNA copies in BM showed a significant drop for mice treated with CD19^KICAR-T cells ( Figure 6G ). All mice treated with gp350^KICAR-T cells showed progressive disease (PD) ( Figures 6C, D ). The persistence of cells expressing CD19⁺ and gp350⁺ within the GFP⁺CD20⁺ blast cell populations recovered from BM and SPL of mice was further assessed by flow cytometry. Interestingly, compared with mice that were only challenged with lymphoma, mice treated with gp350^KICAR-T cells showed slightly lower frequencies of cells expressing gp350 ( Figures 6H, I ). Analyses of CD19 and gp350 on lymphoma cells of mice treated with CD19^KICAR-T cells were not feasible due to their almost complete elimination. The expression of gp350 was reduced on blasts, which could be associated with mechanisms of antigen escape, reducing the putative therapeutic effects of gp350^KICAR-T cells. Flow cytometry analyses of BM and SPL for detection of ^KICAR-T cells in mice challenged with Daudi were attempted twice, but ^KICAR-T cells were barely detectable on terminal analyses (less than 0.1%, data not shown). Currently, we do not have an explanation for this; this phenomenon could be assessed in the future by cross-sectional analyses.

We subsequently developed the Jiyoye in vivo model to assess the effects of the ^KICAR-T cells ( Figure 7A ). BLI analyses after Jiyoye/fLuc-GFP challenge showed lymphoma development, which occurred particularly quickly in female mice ( Figure 7B ). The BLI signal was mostly localized in the bones, LNs, and ovaries ( Figure 7C ). Analyses of BM smears confirmed the presence of cells with typical BL morphology ( Figure 7D ). GFP⁺ blasts were most prevalent in BM and LNs ( Figure 7E ). Lymphoma GFP⁺ blasts expressed CD19 and CD20 (>99%) ( Figure 7F ). Compared to lymphoma cells maintained in culture, only a small fraction of the GFP⁺ cells recovered from LNs or BM showed detectable gp350 expression ( Figure 7G ). The Jiyoye model was then used to test CD19^KICAR-T and gp350^KICAR-T cells in vivo using a similar scheme as previously employed for the Daudi model; 1 × 10^{6 KI}CAR-T cells were applied i.v. per mouse ( Figure 8A ). The cells were used while fresh after production (CAR positivity for CD19^KICAR-T was 25% and for gp350^KICAR-T cells 19%; CD4/CD8 ratios approximately 50/50) ( Figure 8B ). Longitudinal analyses showed a significant reduction in BLI signal for mice treated with CD19^KICAR-T cells compared with the other cohorts ( Figures 8C, D ). Comparing the signals between weeks 2 and 3, for most mice, the lymphoma spread was reduced ( Figures 8C, D ). Challenged and non-treated mice or mice treated with gp350^KICAR-T cells showed major weight loss on week 3 after challenge ( Figure 8E ). At week 3.5, one mouse of the gp350^KICAR-T cohort died, and at this point, the experiment was terminated to reduce the suffering of the mice and to allow harvesting of the tissues. The frequencies of GFP⁺ blasts in PBL, BM, SPL, and LNs were significantly lower for mice treated with CD19^KICAR-T cells compared with the other cohorts ( Figure 8F ). Detection of EBV DNA in BM was significantly lower for mice treated with CD19^KICAR-T cells (***p < 0.001, n = 6) and for the surviving mice treated with gp350^KICAR-T cells (*p < 0.05, n = 5) compared with challenged non-treated mice (n = 6) ( Figure 8G ). Higher frequencies of gp350^KICAR-T cells T cells than CD19^KICAR-T cells were observed in BM and SP ( Figures 8H, I ). While gp350^KICAR-T cells were >90% CD8⁺CAR⁺, the frequencies of CTLs within CD19^KICAR-T were significantly lower, and <10% were CD8⁺CAR⁺ ( Figures 8J ). Remarkably, GFP⁺CD20⁺ Jiyoye blasts recovered from BM of mice treated with CD19^KICAR-T cells showed loss of CD19 expression ( Figure 8K ). Therefore, the Jiyoye BL model confirmed the therapeutic activity of CD19^KICAR-T cells. Further, the stable disease could be explained by the CD19 antigen loss. However, persisting with gp350^KICAR-T cell therapy was associated with a reduction of the EBV DNA load, although this did not produce therapeutic effects, and the persisting ^KICAR-T cells were almost exclusively CD8⁺ cells.

We have previously shown in a humanized mouse model recapitulating EBV⁺ PTLD that two immunologic hallmarks associated with tumor progression are the increased expression of activation/exhaustion markers on CD8⁺ T cells (PD-1, LAG-3, and TIM-3) and accumulation of CD4⁺ T_regs (27) (28). Therefore, we first assessed in vitro whether co-culture of CD19^KICAR-T cells or gp350^KICAR-T cells with the EBV⁺ BL lines at 1:1 E:T ratio would result in upregulation of the activation/exhaustion makers. EBV-negative Nalm-6 acute lymphoblastic leukemia cells expressing CD19 were included as controls. The co-cultures were maintained for 3 days and were analyzed daily for target cell killing, CD4/CD8 ratios, and expression of the exhaustion markers. CD19^KICAR-T cells eliminated the Nalm-6/fLuc-GFP cells, progressively killed the Daudi/fLuc-GFP cells, and were not effective at killing the Jiyoye/fLuc-GFP cells ( Figure 9A ). Under these experimental conditions, gp350^KICAR-T cells did not react against Nalm-6/fLuc-GFP cells, and only reduced the growth of Daudi/fLuc-GFP and Jiyoye/fLuc-GFP cells ( Figure 9B ). At the end of the cultures, CD19^KICAR-T cells showed higher frequencies of CD8⁺ than CD4⁺ T cells, but this trend was not observed for gp350^KICAR-T cells ( Figure 9C ). CD19^KICAR-T cells exposed to Nalm-6 cells showed low expression of the exhaustion markers, similar to the baseline control (without target cells) ( Figure 9D ). CD19^KICAR-T cells incubated with either Daudi or Jiyoye showed similar upregulation of PD-1 on CD4⁺ cells and of PD-1 and LAG-3 on CD8⁺ cells ( Figure 9D , gating strategy details in Figure S6 ). In contrast, gp350^KICAR-T cells exposed to Nalm-6 or to Daudi cells did not show upregulation of exhaustion markers. However, gp350^KICAR-T cells co-cultured with Jiyoye, similarly to the outcome with CD19^KICAR-T cells, showed upregulation of PD-1 and LAG-3 on the CD4⁺ and of LAG-3 on CD8⁺ T cells ( Figure 9E ). Thus, overall, the in vitro data indicated that Jiyoye promoted the exhaustion of both CD19^KICAR-T and gp350^KICAR-T cells. Subsequently, we evaluated the effects of Jiyoye in vivo. We used the cryopreserved BM samples shown in Figure 8F of mice challenged with Jiyoye and treated with ^KICAR-T cells. GFP⁺ blasts were almost undetectable in mice treated with CD19^KICAR-T cells, but were present in mice treated with gp350^KICAR-T cells ( Figure 9F ). On the other hand, compared with CD8⁺CD19^KICAR-T cells, significantly higher expression of PD-1 and LAG-3 was observed in CD8⁺gp350^KICAR-T. Thus, from this series of analyses, we concluded that the short 3-day co-culture of Jiyoye with ^KICAR-T cells promoted their activation and exhaustion in a similar way. However, their longer exposure to Jiyoye in vivo resulted in a dichotomy between higher occurrences of CD4⁺CD19^KICAR-T_regs and exhaustion of CD8⁺gp350^KICAR-T cells (see Table S2 ).

We used BM samples recovered from mice challenged with Jiyoye to detect T_regs by flow cytometry, defined as CD4⁺FoxP3⁺ ( Figure 9G , see gating strategy details in Figure S7 ). As staining controls, we used PBMCs (positive controls, range of 1.01%–12.70% T_regs) or ^KICAR-T cells maintained in culture (approximately 10% T_regs). Mice challenged with Jiyoye and treated with CD19^KICAR-T cells showed significantly higher T_reg frequencies (range of 0.05%–4.25% T_regs) than mice treated with gp350^KICAR-T cells (less than 0.15% T_regs). This was correlated with the previously seen bias toward accumulation of CD4⁺CD19^KICAR-T cells and CD8⁺CD19^KICAR-T cells ( Figure 9H , see Table S2 ). The significantly higher rate of detection of TIM-3⁺CD4⁺ cells within CD19^KICAR-T than gp350^KICAR-T cells represents another interesting finding since TIM-3 expression is a marker expressed on functional T_regs in human tumors (29).