Meditope-Enabled Chimeric Antigen Receptors Facilitate Plug-and-Play Control of T Cells

Treatment of solid tumors with CAR T-cell therapy has proven challenging. The authors report a versatile and modular platform that enables programmable control of CAR T-cell activity, offering a new strategy to overcome tumor heterogeneity and immune suppression.

Abstract

Chimeric antigen receptor (CAR) T cells have transformed cancer treatment, yet challenges for achieving broader clinical success remain, including overcoming tumor antigen heterogeneity and limited T-cell fitness. To address these challenges and enhance CAR T-cell functionality, we leveraged meditope technology, a lock-and-key platform where fragment antigen-binding (Fab) regions of antibodies are modified to bind a small cyclic peptide termed meditope peptide (meP). We developed a panel of meditope-enabled Fab-based CARs (meCAR), which showed selective binding to the meP and comparable activity with traditional single-chain variable fragment–based CARs. Focusing on HER2-targeted meCARs for evaluating platform utility, we exploited the modularity of the meditope platform to detect meCAR T cells using meP-fused fluorescent agents, promote meCAR T-cell expansion via meP-fused IL15 cytokine, and broaden tumor antigen targeting through meP-fused antibodies to address tumor heterogeneity. These findings establish the meditope technology as a versatile strategy to augment CAR T-cell functionality and overcome key limitations of current CAR-based therapies.

See related Spotlight, p. 1536

Introduction

The performance of current chimeric antigen receptor (CAR) T-cell therapies is constrained by fixed features encoded in the CAR design, limiting opportunities for fine-tuned regulation and expanded functionality. These hard-wired CARs have achieved remarkable clinical success in relapsed/refractory B-cell malignancies (1–5), clinical settings in which tumor antigen expression is relatively homogeneous. However, these fixed CAR designs have had limited success against solid tumors due to many challenges, including tumor heterogeneity, limited persistence, and the suppressive tumor microenvironment (TME). Furthermore, unregulated CAR T-cell expansion and cytokine production can result in life-threatening toxicities, including cytokine release syndrome and immune effector cell–associated neurotoxicity syndrome (6, 7). These limitations underscore the need for more advanced CAR designs that enable tunable control and expanded functionality.

We incorporated meditope technology as part of a programmable CAR T-cell system. Meditope technology effectively transforms a unique cyclic peptide [meditope peptide (meP)] and an engineered binding site within the fragment antigen-binding (Fab) region of an antibody (meditope-binding site) into an orthogonal but inert ligand/receptor pair that can be exploited to add functionality to any meditope-enabled Fab (meFab) or monoclonal antibody (mAb). meP is a 12–amino acid cyclic peptide (sequence: CQFDLSTRRLKC) that was originally identified by its binding to the therapeutic antibody cetuximab through a noncovalent interaction without interfering with antigen recognition (8, 9). Based on the crystal structure of the meP–cetuximab complex, distinct residues were identified and grafted onto human mAbs, enabling meP binding without affecting antigen binding affinity (8, 10). Other compounds including fluorophores and biologics, have been genetically or chemically coupled to meP, adding novel functionality to meditope-enabled mAbs (9).

We hypothesized that Fab-derived CARs could be meditope-enabled (hereafter referred to as “meCAR”) to allow binding of meP-fused compounds, offering a modular approach to enable CAR T-cell programmability. As proof of concept, we successfully generated meCAR constructs by replacing the conventional single-chain variable fragment (scFv) domain with an meFab. Furthermore, we utilized meP-fused compounds to visualize meCAR T cells by flow cytometry and immunofluorescence (IF), enhance expansion of meCAR T cells, and establish a dual-targeting meCAR T-cell platform without additional genetic engineering. Our study demonstrates the potential of combining CAR T-cell therapy with meditope technology to overcome limitations of current hard-wired CARs and advance the field of engineered cell therapies.

Materials and Methods

Cell lines

Human breast cancer cell lines MCF7 (ATCC HTB-22, RRID: CVCL_0031), SKBR3 (ATCC HTB-30, RRID: CVCL_0033), OVCAR3 (ATCC HTB-161, RRID: CVCL_0465), MDA468 (ATCC HTB-132, RRID: CVCL_0419), 293T (ATCC CRL-3216, RRID: CVCL_0063), and RL95 (ATCC CRL-1671, RRID: CVCL_0505) were maintained in DMEM (Gibco, cat. #11960051, Thermo Fisher Scientific) supplemented with 10% heat-inactivated FBS (HyClone) and 2 mmol/L L-glutamine (Irvine Scientific). Lymphoblastoid cell lines (LCL) Nalm6 (ATCC CRL-3273, RRID: CVCL_UJ05), K562 (ATCC CCL-243, RRID: CVCL_0004), Jurkat (ATCC TIB-152, RRID: CVCL_0367), and Raji (ATCC CCL-86, RRID: CVCL_0511) were maintained in RPMI (Gibco, cat. #22400089) supplemented with 10% heat-inactivated FBS. To generate tumor cell lines expressing enhanced green fluorescent protein (GFP) and firefly luciferase (ffLuc), MCF7, Raji, OVCAR3, and MDA468 cells were transduced with a lentiviral vector (epHIV7 backbone) encoding GFP-ffLuc under the control of EF1 promoter as described previously (11, 12). After expansion in complete medium (culture conditions described above) for 14 days, GFP⁺ cells were sorted by FACS for >98% purity and expanded for in vivo experiments.

To generate MDA468 cell lines overexpressing target antigens, the coding sequences for the extracellular domain of CD20 (InvivoGen, cat. #puno1-hCD20), HER2 (UniProt: P04626), and PD-L1 (OriGene, cat. #SC115168) were subcloned into the epHIV7 lentiviral vector. Lentiviruses were subsequently used to transduce MDA468 cells to establish stable antigen-expressing cell lines. Cells were sorted and maintained in the mediums described above. All cell lines were routinely tested for Mycoplasma contamination and passaged at least twice before experimental use.

meCAR constructs and lentivirus production

The clinical grade scFv-based HER2-meCAR was generated as previously described, and the Fab region used in the HER2-meCAR construct was obtained from meditope-enabled trastuzumab (8, 12, 13). Briefly, the expression of variable light/constant light (V_L/C_L)and variable heavy/constant heavy (V_H/C_H) chains was mediated by a T2A ribosomal skip mechanism to generate HER2-meCAR-LH or HER2-meCAR-HL, depending on the expression order of V_L/C_L and V_H/C_H (Fig. 1A). Additionally, we utilized the L60 linker to stabilize the pairing of V_L/C_L and V_H/C_H for creating the single-chain format of the HER2-meCAR-SC construct. The CD19 and HER3/EGFR meCARs were created by complementarity-determining region (CDR) grafting from meditope-enabled FMC63 and MEHD7945A mAbs, respectively, to a 4D5 framework (Supplementary Fig. S1A; refs. 1, 14). The extracellular domain included 129 amino acids of hinge, linker, and CH2-deleted IgG4 Fc (CH3), and the intracellular domains were composed of CD28 costimulatory and cytolytic CD3ζ signaling domains (13, 15). 293T cells were transfected with meCAR constructs and lentiviral packaging plasmids using polyethylenimine (PEI) methods. Briefly, lentivirus was produced by seeding 293T cells in T225 tissue culture flasks 1 day prior to transfection. Packaging plasmids and the CAR expression construct were mixed with PEI (1 μg/μL) at a DNA ratio of 1:2 (w/w) and incubated for complex formation before being added dropwise to the cells. Supernatants were collected and followed by high-speed centrifugation to enrich lentivirus. Viral titers were determined using Jurkat cells (ATCC TIB-152, RRID: CVCL_0367) based on Fc expression.

Figure 1.

Engineering and characterization of meCAR T cells targeting HER2. A, Schematic of meCAR constructs, with light chain (V_LC_L) and heavy chain (V_HC_H) sequences in relation to the IgG4-derived CH3 hinge, CD28 transmembrane/costimulatory, and CD3ζ signaling sequences, as well as the T2A ribosomal skip or L60 linker sequences as indicated. meP binding domain in the resulting CAR protein schematic. B, meCAR expression by flow cytometric staining with anti-Fc (left) and meP binding using meP-647 (right). UTD cells were used as the negative control. C and D, Percentage of cells expressing CD107a (C) and intracellular IFNγ (iIFNγ; D) for the indicated HER2-meCAR T-cell variants (gated on viable CAR+ cells) after coculture with HER2-negative MDA468 tumor cells, HER2-positive MCF7 cells, or SKBR3 tumor cells using a one-way ANOVA test. E and F, Kinetics of tumor cell killing performed by xCELLigence assay upon coculture of the indicated meCAR T cells with MCF7 (E) or SKBR3 (F) tumor cells. Mean cell index measurements reflect tumor cell viability of triplicate wells at each time point. Bar charts depict the percent cytolysis as compared with coculture with UTD cells at last measured time point using a one-way ANOVA test. *, P < 0.05; ***, P < 0.001; ****, P < 0.0001. G, xCELLigence killing assays in the presence or absence of 100 nmol/L meP-647. Percent MCF7 (top) or SKBR3 (bottom) cytolysis upon coculture with the indicated meCAR T cells at each time point. H–J, ffLuc⁺ MCF7 cells were intraperitoneally injected into NSG mice. On day 8 after implantation, mice were left untreated or were treated with an intraperitoneal injection of either UTD, meCAR-HL, or meCAR-SC T cells followed by biophotonic imaging for tumor size over time (H and I). H, Mean ± SD tumor flux values for each group using a two-way ANOVA test. J, Kaplan–Meier survival analysis of each group using a Mantel-Cox test. Dashed line indicates the day of T-cell treatment. ns, nonsignificant; ***, P < 0.001; ****, P < 0.0001.

CAR T-cell production

Naïve/memory T (Tn/mem) cells were isolated from healthy blood donors under protocols approved by City of Hope (COH) Institutional Review Board (IRB; ref. 16). Briefly, peripheral blood mononuclear cells were isolated by density gradient centrifugation over Ficoll-Paque (GE Healthcare) and underwent a series of selection by CliniMACS/autoMACS to remove CD14⁺ (Miltenyi Biotec cat. #130-050-201) and CD25⁺ (Miltenyi Biotec cat. #130-092-983) cells, followed by CD62L (Miltenyi Biotec cat. #130-091-758) positive selection for Tn/mem cells. To generate CAR T-cell products, T cells were stimulated with Dynabeads Human T-Expander CD3/CD28 (Invitrogen) at a 1:3 ratio (T-cell:bead) overnight in X-VIVO 15 (Lonza) supplemented with 10% FBS, 2 mmol/L L-glutamine, 50 U/mL IL2 (Novartis NDC 73776-022-01), and 0.5 ng/mL IL15 (CellGenix cat. #1013-050). Stimulated T cells were then transduced with lentivirus (multiplicity of infection: 1) encoding the meCAR. Mock and CAR transduced T cells were cultured with indicated cytokines 3 times a week for 16 to 18 days before subsequent analyses.

Fluorescence quantitation using quantum molecules of equivalent soluble fluorochrome beads

Quantitative flow cytometry was performed using quantum molecules of equivalent soluble fluorochrome (MESF) microsphere kits (Bangs Laboratories cat. #647A) according to the manufacturer’s instructions. On the day of analysis, MESF beads and Fc-stained cell samples (Jackson ImmunoResearch cat. #109-606-008) were run on the same flow cytometer using identical photomultiplier tube (PMT) voltages and compensation settings to generate a fluorescence calibration curve. Bead suspensions were prepared immediately prior to use and protected from light. Singlet bead populations were gated by forward scatter (FSC)/side scatter (SSC), and the median fluorescence intensity values were recorded for each population. Fluorescence intensities of stained cell samples were then converted to MESF units based on the calibration curve, enabling standardized and quantitative comparison of antigen expression across experiments.

Animal studies

All mouse experiments were approved by the COH Institutional Animal Care and Use Committee (IACUC protocol #18059). For MCF7 and OVCAR3 models, 8-week-old female NOD/SCID gamma (NSG) mice (The Jackson Laboratory) were intraperitoneally injected with 5 × 10⁶ of ffLuc⁺ tumor cells for 7 days, followed by randomized grouping for intraperitoneal treatments with 2 × 10⁶ of untransduced (UTD) T cells or CAR T cells on day 8. For redirected- and dual-targeting models, the NSG mice were intraperitoneally injected with ffLuc⁺ MDA468-CD20 or mixture MDA468-CD20/MDA468-HER2 tumor cells, as indicated in the experimental schema for 4 days, followed by randomized grouping for intraperitoneal treatments with 5 × 10⁶ of UTD or meCAR T cells. Mice were then treated with meP–anti-CD20 at a dose of 1 mg/kg every 2 days for a total of 10 administrations. Tumor burden was imaged at the indicated time points with SPECTRAL LagoX (Spectral Instruments Imaging) and analyzed using Aura software (v2.3.1, Spectral Instruments Imaging); measurements <10⁶ photons/second represent background flux levels. Survival curves were generated using GraphPad Prism software. Tissues were harvested for evaluation of tumor antigen expression using a MACSQuant Analyzer (Miltenyi Biotec) in end-stage mice occurred after euthanasia.

For cytokine treatment studies in the NSG mouse model, mice received an intraperitoneal injection of 2 × 10⁶ meCAR T cells, followed by daily intraperitoneal administration of IL2 (Novartis NDC 73776-022-01) or meP-fused IL15 superagonist (meP-IL15s; 1 nmol/mouse) for 10 consecutive days. Blood and tissue samples were collected at the indicated time points for downstream analyses.

Generation of meP-fused compounds

The IL15 superagonist (IL15s), comprising the IL15 receptor α (IL15Rα) sushi domain and IL15, was cloned into a mammalian expression vector encoding a human IgG1 Fc fusion protein with a tobacco etch virus (TEV) protease cleavage site to generate the Fc-IL15s construct. A meditope sequence was subsequently inserted between the Fc domain and the IL15s to generate the Fc-meP-IL15s construct. Both constructs were verified by DNA sequencing prior to expression. Fc-IL15s and Fc-meP-IL15s fusion were transiently expressed in a Chinese hamster ovary (CHO) cell line using the ExpiCHO system (Thermo Fisher Scientific) according to the manufacturer’s manual. The culture supernatant was collected on day 7 and purified through affinity chromatography (ProG resin, GenScript) followed by size-exclusion chromatography (Cytiva Lifesciences). To generate IL15s and meP-IL15s, Fc was cleaved through the TEV cleavage site between the Fc and IL15s with and without meP and removed through reverse affinity chromatography. Biophysical validation of those molecules was performed via surface plasmon resonance (SPR; Biacore T200, Cytiva Lifesciences). Briefly, human IL2 receptor β/γ heterodimer protein (ACROBiosystems cat. #ILG-H5254) was immobilized on a CM5 sensor chip via amine coupling using the amine coupling kit (Cytiva Lifesciences cat. #BR100050), and human IL15 (Abcam cat. #ab259403) was included as a control.

Anti-CD20 (rituximab) and its meP-fused variants were generated by fusing the meditope to either the N-terminus of the heavy chain (HCN) or N-terminus of the light chain (LCN). Mouse serum albumin (MSA) binding peptide (RLIEDICLPRWGCLWEDD) was fused to the C-terminus of the antibody light chain (17). Each construct was produced in ExpiCHO-S cells (Thermo Fisher Scientific) transiently. Following transfection with the mammalian expression vectors (ATUM cat. #pD2610-v13) encoding each construct using ExpiFectamine (Thermo Fisher Scientific), cell cultures were incubated in a 37°C shaking incubator with >80% humidity and 8% CO₂ for 1 day and then moved to a 32°C shaking incubator with 5% CO₂ for the next 6 days. Purification was done with the appropriate affinity and size-exclusion chromatography. Binding affinity for antigen and meditope-enabled antibody was measured on Biacore T200 (Cytiva Lifesciences). Kinetic characterization was determined using Biacore evaluation software. Briefly, human CD20 (ACROBiosystems cat. #CD0-H52H3) or MSA (MilliporeSigma cat. #3559) was immobilized on a CM5 chip via amine coupling using the amine coupling kit (Cytiva Life Sciences cat. #BR100050). An antibody without the fusion peptide served as a control.

The meP-fused anti–PD-L1 (atezolizumab) was generated and characterized using the same approach described for the meP-fused anti-CD20. Antibodies were transiently expressed in ExpiCHO-S cells (Thermo Fisher Scientific) and purified prior to analysis. For binding studies, recombinant human PD-L1 (UniProt: Q9NZQ7; amino acids 19–238), meP, or MSA (MilliporeSigma, cat. #3559) was immobilized on the SPR sensor chip. The antibody lacking fusion peptide was included as a control.

Flow cytometry and data analysis

Expression of meCARs on T cells was analyzed with meP-fused Alexa Fluor 647 (meP-647; COH) or antibodies against human Fc (Jackson ImmunoResearch cat. #109-096-008 and 109-606-008), and clinical scFv CAR expression was determined by staining truncated CD19 (BD Biosciences cat. #557835, clone SJ25C1). T cells were characterized using antibodies against CD4 (BD Biosciences cat. #347324, clone SK3), CD8 (BD Biosciences cat. #347314, clone SK1), CD45RA (BD Biosciences cat. #563031, clone H100), and CD62L (BD Biosciences cat. #559772, clone DREG56). T-cell activation was determined using antibodies against CD107a (BD Biosciences cat. #555800, clone H4A3), 4-1BB (BD Biosciences cat. #555956, clone 4B4-1) and IFNγ (BD Biosciences cat. #557643, clone B27). Antibodies for pSTAT5 (BD Biosciences cat. #612567 clone 47), and Ki67 (BioLegend cat. #350504 clone Ki67) were used to analyze preferential meCAR expansion by meP-IL15s. Tumor antigen expression was analyzed using antibodies against HER2 (Jackson ImmunoResearch, cat. #FAB9589P-100, clone Hu5), PD-L1 (BioLegend, cat. #374506, clone MIH3), EGFR (BioLegend, cat. #352910, clone AY13), and HER3 (BioLegend, cat. #324706, clone 1B4C3).

All samples were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (Thermo Fisher Scientific cat. #D21490), acquired on a MACSQuant analyzer (Miltenyi Biotec), and analyzed using FCS Express 7 (version 7.30.0018). Debris was excluded based on FSC and SSC characteristics, followed by singlet isolation using FSC-A versus FSC-H. Dead cells were excluded using DAPI staining. Live single cells were subsequently gated on the markers described above and quantified within the gated populations.

In vitro T-cell functional assays

Tumor cells were cultured in the medium described above, and meCAR T cells were washed/resuspended in the same medium followed by coculturing with tumor cells at the indicated effector:tumor (E:T) ratios. For degranulation and intracellular IFNγ assays, meCAR and tumor cells were incubated at an E:T of 1:1 (20,000:20,000) for 5 hours with the addition of anti-CD107a (BD Biosciences cat. #555800 clone H4A3) and GolgiStop protein transport inhibitor (BD Biosciences cat. #554724). After the coculture, cells were harvested, fixed, permeabilized (BD Biosciences cat. #554714), and analyzed by flow cytometry as described above. To test cytolytic activity, tumor and CAR⁺ T cells were cocultured at indicated E:T ratios for 48 hours, and the killing was quantified by flow cytometry analyzing viable tumor cells (GFP⁺/DAPI⁻). All experiments were performed with ≥3 biological replicates.

In situ IF staining and image analysis

Primary brain tumor (PBT) PBT103-2 was obtained from a glioblastoma (GBM) patient resection at COH under written consent and protocols approved by the COH IRB and in accordance with ethical standards (18, 19). A total of 1 × 10⁶ tumor cells were subcutaneously engrafted in NSG mice. Tumor volume was monitored using digital caliper measurements and calculated using the following formula: tumor volume (mm³) = (L × W²)/2, where L is the longest tumor diameter and W is the shortest tumor diameter. Once tumor volume reached 100 mm³, mice were intratumorally treated with 2 × 10⁶ of scFv or meCAR T cells. After 24 hours, GBM tumors were frozen sectioned and fixed in 4% paraformaldehyde for 15 minutes. After fixation, samples were blocked in PBS containing 0.25% BSA and 0.3% Triton X-100 for 1 hour. Samples were stained with rabbit anti–human Her2 (Agilent A048529-1) and mouse anti–human CD3 (Agilent M725429-2) primary antibodies for 1 hour followed by anti–rabbit Alexa 488 (Thermo Fisher Scientific A-21206), anti–mouse Alexa 555 (Thermo Fisher Scientific A-31570), and meP-647 staining for 30 minutes. Finally, samples were counterstained by DAPI. The images were acquired by ZEISS Axio Observer and analyzed using ZEISS ZEN software (version 3.13).

Cytokine profiling

HER2-meCAR T cells (20,000) were cultured with Raji tumor cells (E:T = 1:1) in the absence or presence of either meP-fused anti-CD20 IgG (meP–anti-CD20-IgG) or control anti-CD20 IgG (100 nmol/L) for 48 hours. Harvested supernatants were analyzed for GM-CSF, IFNγ, IL2 and TNFα by cytometric bead array using the Human Cytokine 10-Plex Panel (Thermo Fisher Scientific) according to the manufacturer’s instructions.

ELISA assay of mouse serum meP-IL15s

The ELISA kits of meP-IL15s were purchased from R&D Systems (cat. #DY6924). Mouse serum samples were collected at the 3, 6, and 18 hours after a single administration of 1 nmol meP-IL15s/mouse. Serum samples were diluted 1:10 in diluent provided by kit and analyzed according to the manufacturer’s instructions. The microplates were analyzed using a Cytation 5 (Agilent) plate reader.

xCELLigence killing assays

Real-time cytotoxicity of HER2-meCAR T cells was evaluated using the xCELLigence Real-Time Cell Analysis (RTCA) system (Agilent Technologies) against MCF7 or SKBR3 tumor cells according to the manufacturer’s instructions. Briefly, 25,000 tumor cells were seeded per well and cultured overnight before the addition of HER2-meCAR T cells at an E:T ratio of 1:8, either alone or in the presence of meP-647. Cell growth was monitored continuously using the xCELLigence RTCA system, with impedance measurements recorded every hour. Redirected targeting and dual-targeting assays using meP-fused antibodies were performed following the same procedure.

Statistical analysis

Data were presented as mean ± SD and statistical comparisons analyzed using the one-way or two-way ANOVA for independent or grouped samples to calculate P values, unless otherwise stated. *, P < 0.05; ns, not significant. Comparison of Kaplan–Meier survival data was performed using Prism v9.0 (GraphPad Software).

Results

Development of the meCAR T-cell platform

Unlike traditional CAR constructs that target via an scFv, engineering the meP-binding site required construction of Fab-based CARs (Fig. 1A). We optimized Fab-CAR design using a trastuzumab-based HER2-targeted CAR based on prior studies establishing the amino acid substitutions conferring meP binding (Supplementary Fig. S1A; ref. 8) and our experience with HER2-CAR T cells (12). We optimized the HER2-meCAR design, generating three constructs that varied in the meFab structural framework (Fig. 1A): (i) meCAR-LH (Fab light chain CAR, followed by T2A and Fab heavy chain); (ii) meCAR-HL (Fab heavy chain CAR, followed by T2A and Fab light chain); and (iii) meCAR-SC [single-chain (SC) Fab CAR; ref. 20]. All three meCARs were successfully expressed as detected by anti-Fc (Fig. 1B). The binding of meP-647 to meCAR T cells confirmed proper Fab folding and retention of the meP-binding site (Fig. 1B). Moreover, the utility of the L60 linker in stabilizing the pairing of heavy and light chains was supported by the fact that the meCAR-SC construct displayed the highest CAR expression and meP binding (Fig. 1B), as well as the strongest HER2 antigen binding (EC₅₀ < 1 nmol/L; Supplementary Fig. S1B).

We examined effector function of HER2-meCAR T cells by measuring CD107a degranulation and IFNγ secretion upon coculture with MCF7 and SKBR3 breast cancer cell lines, which exhibit moderate and high HER2 expression (Supplementary Fig. S2A), respectively. We observed comparable activity for all meCAR T-cell variants (Fig. 1C and D). In xCELLigence-based killing assays, HER2-meCAR T cells killed both MCF7 (Fig. 1E) and SKBR3 (Fig. 1F), with the meCAR-SC construct exhibiting superior killing of SKBR3 (Fig. 1F). The addition of meP-647 did not affect the ability of meCAR T cells to kill either cell line (Fig. 1G), confirming that binding of meP to meFab did not interfere with meCAR antigen recognition (8). We compared in vivo antitumor activity of HER2-meCAR T cells, testing only the meCAR-HL and meCAR-SC constructs, which showed the highest meCAR expression (Fig. 1B) and in vitro functionality (Fig. 1F). Using an MCF7 tumor xenograft model in NSG mice, we observed potent antitumor activity in tumor-bearing mice treated with meCAR-HL or meCAR-SC compared with untreated or UTD T cell–treated controls (Fig. 1H and I). The reduction of bioluminescence signals in both untreated and UTD groups over time was likely attributable to solid tumor dissemination throughout the peritoneal cavity, leading to low penetration of luciferin substrate, as visibly extended abdomens with high peritoneal tumor burden upon euthanasia were clearly observed. Nevertheless, the meCAR-SC T cells eliminated tumor in all mice (N = 6) and prolonged survival for >80 days, modestly outperforming meCAR-HL (Fig. 1H–J).

To investigate expression and functional differences between Fab-based CARs versus standard scFv-based CARs, we compared the preferred HER2-meCAR-SC design with our clinical scFv-based HER2-CAR (12). The flow cytometric mean fluorescence intensity of meCAR expression and absolute meCAR molecules per cell were found to be comparable with that of the scFv-based HER2-CAR (Supplementary Fig. S2B). Moreover, meCAR-SC demonstrated antitumor efficacy comparable with our clinical scFv-based HER2-CAR in a HER2⁺ ovarian cancer xenograft model, achieving similar tumor control and survival benefit in vivo, with no evidence of tumor in the remaining mice of each CAR T-cell treatment group (n = 5) at day 125 (Supplementary Fig. S2C–S2E). Together, these data suggest that Fab-CARs can be generated with comparable expression and functional potency as compared with more standard scFv-CARs.

To evaluate applicability of the meCAR platform to other targets, we incorporated the SC-Fab framework into two other meCARs: (i) a humanized CD19-CAR recently developed by our group (21) and (ii) a HER3/EGFR dual-targeting CAR based on duligotuzumab (MEHD7945A; ref. 14). We confirmed meCAR-SC expression in T cells, binding of meP, and antigen-dependent CAR effector function (Supplementary Fig. S3). These results suggest that meCAR is a generalizable platform.

Overall, these findings demonstrated feasibility of generating Fab-based meCARs that bind meP without compromising CAR-mediated antigen binding and identified the meCAR-SC construct as the preferred meCAR design. Therefore, we moved forward with evaluating the utility of the meCAR platform, focusing on HER2-meCAR-SC as proof of concept.

In situ detection of meCAR T cells

We evaluated the utility of the meditope technology for CAR T-cell detection using fluorescent-fused mePs as tools for visualizing meCAR T cells by fluorescence microscopy. We stained meCAR T cells with anti-Fc or meP-647 reagents and showed strong concordance between meCAR expression (anti-Fc) and meP-647 fluorescence. Furthermore, meP-647 did not stain scFv-based CAR T cells, confirming the specificity of meP (Fig. 2A).

Figure 2.

Immunofluorescent detection of meCAR T cells using meP-fused fluorophores. A, IF staining of UTD T cells, HER2 scFv-based HER2-CAR T cells, or HER2-meCAR T cells (meCAR-HL or meCAR-SC) stained with anti-CD3, anti-Fc, and meP-647. B, In situ detection of HER2-meCAR T cells in tumor tissue. Subcutaneous GFP⁺ tumors were injected with either HER2 scFv-based HER2-CAR T cells or HER2-meCAR-HL T cells, and tumors were harvested, fixed, and processed for IF as in (A).

Detection of CAR T cells in patient tissue samples can be technically challenging. For example, although RNAscope offers high spatial sensitivity by targeting unique transgene sequences like the scFv or untranslated regions, it requires a complex hybridization protocol using customized oligonucleotide probes for CAR constructs (22). In contrast, the meP-647 system utilizes a universal, ready-to-use fluorophore that binds directly to the engineered meditope-binding site on the meCAR protein. Thus, we tested whether meP-647 could be used as a tool to detect meCAR T cells in tumor specimens. We engrafted mice with subcutaneous GBM tumor xenografts, intratumorally injected HER2-meCAR or scFv-CAR T cells, and harvested tumors after 24 hours for analysis. We confirmed the presence of T cells by anti-CD3 IF and specifically identified meCAR T cells using meP-647, which co-localized with CD3 staining in tissues treated with meCAR but not scFv-CAR T cells (Fig. 2B). Overall, these experiments demonstrated that meP-fused compounds offer a universal strategy for detecting meCAR T cells in situ and thus could facilitate detection of meCAR T cells in clinical tissue samples.

Selective expansion of meCAR T cells by meP-fused IL15 cytokine

CAR T-cell persistence and expansion following adoptive transfer are critical determinants of therapeutic response (23). These properties can be enhanced by providing γc cytokines, such as IL2 or IL15, either through exogenous delivery or cell engineering (24–26). IL15 has been shown to support a memory phenotype and enhance CAR T-cell function (27, 28), prompting us to investigate an meP-fused IL15 cytokine for selectively regulating meCAR T-cell expansion and persistence. IL15 itself has moderate binding affinity to its β/γ chain signaling receptor, and its binding and stability are enhanced when fused to the IL15Rα sushi domain to generate a chimeric IL15:sushi superagonist (IL15s; ref. 29). We therefore engineered meP-IL15s to evaluate its effects on meCAR T cells (Fig. 3A).

Figure 3.

meP-IL15s selectively drives meCAR T-cell expansion in vitro. A, Schematic of meP-IL15s interaction with meCAR and IL15 receptor. B and C, HER2-meCAR T cells were cultured in IL2 (50 IU/mL), IL15s (20 ng/mL), or meP-IL15s (20 ng/mL) and monitored for expansion (B) and the percentage of meCAR⁺ T cells (C) over 9 days. C, Representative density plots of DAPI⁻ gated cells stained with anti-Fc to detect meCAR⁺ cells (left), with the percentage of meCAR⁺ cells under each culture condition over time (right). D and E, HER2-meCAR T cells cultured in IL2, IL15s, or meP-IL15s for 5 days were stained for CAR expression (anti-Fc) and intracellular pSTAT5 (D) or Ki67 (E). F and G, Mean percentage of viable UTD or meCAR⁺ T cells that were pSTAT5⁺ (F) and Ki67⁺ (G) after culture with IL2, IL15s, or meP-IL15s (as in the plots of D and E). ns, nonsignificant; *, P < 0.05; ****, P < 0.0001 (using a two-way ANOVA test). H, Percentages of viable meCAR⁺ T cells after culture for 5 days with IL2, IL15s, or meP-IL15s (as in the histograms of D and E). ***, P < 0.001 (using a one-way ANOVA test). I and J, T cells were cultured for 18 days with cytokine supplementation and media replacement every 2 days, followed by flow cytometric analysis of CD4:CD8 ratios (I) and central memory phenotype, defined as CD62L⁺:CD45RA⁻ (J). *, P < 0.05 (using a one-way ANOVA test). Tcm, central memory T.

We cultured HER2-meCAR T cells (45% CAR⁺) with meP-IL15s, IL2, or IL15s and observed comparable expansion (≥15-fold by day 9; Fig. 3B). meCAR⁺ T-cell frequency markedly increased with meP-IL15s, increasing from 57% on day 5 to 92% on day 9, whereas the percentage of meCAR⁺ T cells cultured in IL2 and IL15s remained stable (45%–53%; Fig. 3C). This enrichment of meCAR⁺ T cells by IL15s required the presence of the meP (Supplementary Fig. S4A and S4B). Mechanistically, preferential activation of pSTAT5 and Ki67 was observed in meCAR⁺ T cells cultured with meP-IL15s (Fig. 3D and E). In contrast, pSTAT5 and Ki67 levels were comparable between UTD and meCAR T cells under no cytokine, IL2, or IL15s conditions (Fig. 3F and G). We further observed that meP-IL15s significantly enhanced pSTAT5 and Ki67 expression in meCAR⁺ T cells and drove their selective expansion within 5 days (from 18%–66% meCAR⁺), whereas other conditions showed no enrichment (Fig. 3H). To determine whether T cells undergo phenotypic changes under IL2, IL15s, or meP-IL15s culture conditions, we analyzed the CD4⁺/CD8⁺ ratio and central memory phenotype (defined as CD62L⁺CD45RA⁻) on day 18 following extended culture (28). We observed no significant differences in the CD4⁺/CD8⁺ ratio across all conditions (Fig. 3I). In contrast, IL15s more effectively preserved a central memory phenotype compared with IL2, independent of the presence of the meP (Fig. 3J). Overall, these data demonstrate that meP-IL15s supports selective expansion of meCAR T cells through cis presentation to the IL15 receptor, resulting in pSTAT5 activation. Furthermore, these findings suggest that meP-IL15s supports expansion without altering the CD4⁺/CD8⁺ ratio and maintains a central memory phenotype comparable with IL15s conditions.

We next evaluated the utility of IL15s for selectively supporting meCAR T-cell persistence and expansion in vivo. For these experiments, we administered HER2-meCAR T cells into the peritoneal cavity of NSG mice followed by 10 daily intraperitoneal injections of either IL2 or meP-IL15s (1 nmol/mouse) and analyzed the tissue samples on day 9 after treatment (Fig. 4A). Treatment with meP-IL15s led to a significant increase in the percentage of meCAR⁺ T cells in ascites (94%), blood (93%), and spleen (79%; Fig. 4B and C) along with a significant increase in total meCAR⁺ T-cell numbers across all 3 tissues (Fig. 4B and D). In contrast, control and IL2-treated mice maintained similar levels of meCAR⁺ T cells. This meP-IL15s–driven enrichment of meCAR⁺ T cells was detected in the ascites as early as day 6 of treatment, indicating a rapid and selective expansion of meCAR T cells (Supplementary Fig. S4C).

Figure 4.

meP-IL15s selectively drives meCAR T-cell expansion in vivo. A, Schematic of meP-IL15s and HER2-meCAR T-cell treatments in vivo. B–D, Tissue samples of ascites, blood, and spleen from NSG mice were examined by flow cytometry for the presence of HER2-meCAR⁺ T cells. Representative dot plots of CD3-gated cells stained with anti-Fc to detect meCAR⁺ cells (B). Tissue samples of ascites, blood, and spleen were analyzed for the enrichment (C) and overall number (D) of HER2-meCAR T cells. E, Schematic of meP-IL15s dose titration in vivo and samples analyzed on day 14. F–H, Ascites samples from NSG mice were examined by flow cytometry (F). The enrichment (G) and overall CD3⁺ T cells (H) were analyzed at indicated meP-IL15s dosage. I and J, The HER2-meCAR T cells were analyzed by meP-647⁺ cells (I), and enrichment (J) and overall CD3⁺ T cells (K) were analyzed at indicated meP-IL15s dosage. ns, nonsignificant; *, P < 0.05; **, P < 0.01; ***, P < 0.001 (using a one-way ANOVA test). Bars depict mean ± SD of the percent cells and numbers of meCAR⁺ cells in each sample.

To investigate the sensitivity of meCAR T cells to meP-IL15s, we performed a dose titration (0.02–1 nmol/mouse) in NSG mice (Fig. 4E). We found that the percentage (Fig. 4F and G) and the overall number of CD3⁺ T cells (Fig. 4F and H) increased at a high (1 nmol/mouse) meP-IL15 dose. The lowest dose of meP-IL15s (0.02 nmol/mouse) led to a modest enrichment in meCAR T-cell levels from 46% under the PBS condition to 69% meCAR positivity (Fig. 4I–K), suggesting that meP-IL15s can selectively enhance the persistence of meCAR⁺ T cells. The intermediate (0.15 nmol/mouse) and high (1 nmol/mouse) meP-IL15s dose levels significantly increased meCAR⁺ T-cell levels in a dose-dependent manner (Fig. 4I–K). This programmable control is especially important in engineered T-cell therapies, in which sustained cellular activity correlates with therapeutic efficacy. Together, these data suggest that the meditope platform allows for selective and fine-tuned modulation of meCAR T-cell persistence and proliferation in vivo with post-infusion administration of meP-fused cytokines, such as IL15s.

To evaluate whether meP-IL15s enhances the antitumor efficacy of meCAR T cells, we used an endometrial tumor model in which RL95 cells were engrafted intraperitoneally, followed by treatment with HER2-meCAR T cells in the absence or presence of control IL15s or meP-IL15s (Fig. 5A). HER2-meCAR T-cell treatment reduced tumor burden by day 14 across all groups, regardless of IL15s or meP-IL15s administration. Mice receiving HER2-meCAR T cells in combination with meP-IL15s exhibited sustained tumor burden reduction (Fig. 5B and C). Consistent with our previous findings, this group also showed a significant in vivo expansion in circulating T-cell numbers (Fig. 5D), suggesting that meP-IL15s enhances the efficacy of HER2-meCAR T cells.

Figure 5.

meP-IL15s enhance meCAR T-cell efficacy in vivo. A–D, ffLuc⁺ RL95 cells were intraperitoneally injected in NSG mice (5 × 10⁶ cells/mice). On day 7 after implantation, mice were left untreated or treated with an intraperitoneal injection of HER2-meCAR T cells (2.5 × 10⁵ cells/mouse), with some groups also receiving 10 daily intraperitoneal injections of either IL15s or meP-IL15s. A, Schematic of the treatment. B and C, Tumor size was followed over time by biophotonic imaging. Mean ± SD tumor flux values for each group (B). *, P < 0.05 (using a one-way ANOVA test). C, Representative mouse images for each group. D, Mean ± SD number of human T cells (CD45⁺CD3⁺) in the blood harvested at end of meP-IL15s were determined by flow cytometry. *, P < 0.05 (using a one-way ANOVA test). E–I, ffLuc⁺ Nalm6 cells were intravenously injected in NSG mice (0.5 × 10⁶ cells/mice). On day 4 after implantation, mice were left untreated or treated with an intravenous injection of either CD19-scFv-CAR or CD19-meCAR T cells (2 × 10⁶ cells/mouse), with some groups also receiving 10 daily intraperitoneal injections of either IL15s or meP-IL15s. E, Schematic of the treatment. F and G, Tumor size was followed over time by biophotonic imaging. F, Mean ± SD tumor flux value for each group. *, P < 0.05 (using a one-way ANOVA test). G, Representative mouse images for each group. H, Kaplan–Meier survival analysis using a log-rank (Mantel–Cox) test. I, Mean ± SD numbers of human T cells (CD45⁺CD3⁺) and CAR⁺ T cells in the blood harvested at day 14 were determined by flow cytometry. *, P < 0.05; **, P < 0.01 (using a two-way ANOVA test).

As a second tumor model, we evaluated CD19-targeted CAR T cells in a CD19⁺ Nalm6 leukemia model under subtherapeutic conditions. Mice were engrafted with Nalm6 tumor cells and treated with CD19-meCAR T cells or CD19-scFv-CAR, with or without IL15s or meP-IL15s (Fig. 5E). Treatment with either CD19-meCAR T cells or CD19-scFv-CAR alone conferred an antitumor response (Fig. 5F–H); however, supplementation of CD19-scFv-CAR T cells with either IL15s or meP-IL15s and supplementation of CD19-meCAR T cells with IL15s did not further extend survival in these groups (Fig. 5H). In contrast, the mice treated with CD19-meCAR T cells in combination with meP-IL15s demonstrated a significant improvement in survival and reduced tumor burden when compared with meCAR control groups with or without IL15s (Fig. 5F–H). Consistent with this, the CD19-meCAR T cells exhibited the greatest in vivo expansion in the meP-IL15s–treated group (Fig. 5I). Given a prior report of IL15s-associated toxicity in vivo (30), we carefully selected a dosing regimen that showed no adverse effects on food consumption or body weight in treated mice (Supplementary Fig. S4D and S4E). Despite the short half-life (detectable in the first 3 hours after administration) of meP-IL15s in the serum (Supplementary Fig. S4F), our findings indicate that meP-IL15s selectively enhances the persistence and therapeutic efficacy of meCAR T cells in vivo, supporting the specificity and functional advantage of this modular cytokine delivery strategy. Efforts to further improve its stability and extend its half-life are currently underway.

Targeting of heterogeneous tumors using meCAR T cells and meP-fused antibodies

Tumor antigen heterogeneity remains a major barrier to effective CAR T-cell therapy (31). We hypothesized that the meCAR platform could serve as a universal docking site for meP-fused antibodies, enabling flexible, programmable multiantigen targeting (Fig. 6A). To test this, we produced meP-fused anti-CD20 nanobody (meP–anti-CD20-Nano) and assessed the activity of HER2-meCAR T cells upon coculture with K562 cells (CD20⁻) or LCL cells (CD20⁺) with or without meP–anti-CD20-Nano. Although HER2-meCAR T cells did not recognize HER2-negative K562 and LCL cells, the addition of meP–anti-CD20-Nano induced dose-dependent degranulation as evidenced by CD107a expression on HER2-meCAR T cells cocultured with CD20⁺ LCL cells but not CD20⁻ K562 cells (Supplementary Fig. S5A). Moreover, meP–anti-CD20-Nano redirected HER2-meCAR T-cell activation (Fig. 6B) and killing (Fig. 6C) when cultured with LCL cells in a dose-dependent manner. We did not observe activation or killing by scFv-based HER2-CAR T cells with the meP–anti-CD20-Nano, indicating that the effects were meditope dependent (Fig. 6B and C). Similarly, meP-fused rituximab (meP–anti-CD20-IgG) stimulated HER2-meCAR T cells to produce pro-inflammatory cytokines (GM-CSF, IFNγ, IL2, and TNFα) when cocultured with CD20⁺ Raji tumors but not in the control cultures with UTD T cells or rituximab (Fig. 6D; Supplementary Fig. S5B). These results demonstrate that meP-fused antibodies can effectively and specifically redirect the effector function of meCAR T cells toward alternative tumor antigens, without compromising the recognition of the original CAR target.

Figure 6.

Redirected targeting of HER2-meCAR T cells mediated by meP–anti-CD20 (αCD20). A, Schematic of meP–αCD20-Nano interaction with meCAR. B and C, HER2-meCAR T cells (20,000) were cultured with either CD20⁻ K562 cells or CD20⁺ LCL cells (E:T = 1:1) in the absence or presence of titrated doses of meP–αCD20-Nano and evaluated by flow cytometry for 4-1BB expression (B) and tumor cell killing (C). Mean ± SD of the percentage of the viable meCAR⁺ gated population (B) or the viable tumor cell–gated population (C) of triplicate wells. scFv-based HER2-CAR T cells and CD20⁻ K562 cells served as controls. D, Cytokine production of HER2-meCAR T cells cultured with Raji tumor cells with or without meP-αCD20-IgG or control αCD20 IgG. UTD T cells were used as the negative control. ****, P < 0.0001 (using a one-way ANOVA test). E–G, Optimization and functional characterization of meP-αCD20 variants. Right in (E), variant schematics. HER2-meCAR T cells were cultured with CD20⁺ Raji cells with or without meP-αCD20-Nano, Fab, and IgG variants for 48 hours followed by flow cytometric analysis of 4-1BB expression (E and F) and tumor cell killing (G). E, Representative density plots of 4-1BB expression profiles of CD3-gated cells under each culture condition. Mean ± SD of the percentage of 4-1BB⁺ meCAR⁺ cells (F) or tumor killing (G) for each group. ns, nonsignificant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 (using a two-way ANOVA test). H and I, Mixture of MDA468-CD20 and MDA468–PD-L1 tumor cells (25,000 cells each) cultured with HER2-meCAR T cells (E:T = 1:8) alone or in the presence of meP–anti–PD-L1 (αPD-L1), meP-αCD20, or both (100 nmol/L) in xCELLigence killing assays (H). ***, P < 0.001; ****, P < 0.0001 (using a one-way ANOVA test). Residual tumor cells were assessed for antigen expression (I).

As meP antibodies can bind to either tumor antigens or meCAR T cells, we explored the mechanism of redirection by preloading HER2-meCAR T cells or LCL tumor cells with meP–anti-CD20-Nano followed by washing 3 times before coculture with the unlabeled counterpart cells (Supplementary Fig. S5C). We observed a more significant increase in CD107a expression on meCAR T cells in the tumor-bound (77%) compared with the T cell–bound condition (16%; Supplementary Fig. S5D), suggesting that redirected targeting of meCAR T cells by meP antibodies occurs primarily through pre-binding of meP antibodies to tumor cells, which are then recognized by meCAR T cells. The presence of meP–anti-CD20-Nano did not impair the cytotoxic activity of HER2-meCAR T cells (Supplementary Fig. S5E).

We next compared antibody formats for meP–anti-CD20, testing nanobody, Fab, and IgG variants with meP fused at the HCN or LCN (Fig. 6E). All formats redirected HER2-meCAR T-cell 4-1BB activation (Fig. 6E and F) and killing (Fig. 6G) against CD20⁺ Raji cells, although the meP-Fab-L variant was less effective at low concentrations (Fig. 6E–G). Redirected meCAR T-cell activation and killing were meP dependent, as they were not observed using the control CD20 antibody lacking the meP (rituximab; Fig. 6G). These findings demonstrate the versatility of the meCAR platform in accommodating diverse antibody formats, including nanobody, Fab, and IgG, for tumor antigen redirection. The results also suggest that HCN fusion is optimal for redirection and will be the default format unless otherwise noted.

Having established that meP-fused antibodies can effectively redirect meCAR T-cell effector function through tumor-bound engagement (Supplementary Fig. S5B and S5C), we next asked whether this modular strategy could be extended to target immunoregulatory molecules, such as PD-L1. To this end, we generated an meP–anti–PD-L1 and evaluated its ability to engage MDA468–PD-L1–expressing tumor cells (Supplementary Fig. S6A). HER2-meCAR T cells were then cocultured with a mixture of MDA468-HER2 and MDA468–PD-L1 tumor cells, either alone or in the presence of the meP–anti–PD-L1. In xCELLigence-based cytotoxicity assays, HER2-meCAR T cells alone mediated only partial tumor killing, whereas the addition of meP–anti–PD-L1 resulted in complete elimination of the mixed tumor culture (Supplementary Fig. S6B).

To further evaluate whether meCAR T cells could support polyclonal antibody–mediated responses, HER2-meCAR T cells were cocultured with a mixture of MDA468-CD20 and MDA468–PD-L1 tumor cells in the absence or presence of meP–anti–PD-L1 and/or meP–anti-CD20. HER2-meCAR T cells alone exhibited no cytotoxic activity against the mixed tumor population, and only partial tumor killing was observed in the presence of either single meP antibody (Fig. 6H). The redirected targeting mediated by meP antibody was antigen specific, as the remaining cells consisted of non-target tumor cells (Fig. 6I). Strikingly, simultaneous administration of both meP–anti-PD-L1 and meP–anti-CD20 resulted in complete tumor eradication, demonstrating that meCAR T cells can mediate a polyclonal response through engagement of multiple meP-fused antibodies (Fig. 6H). These effects are unlikely due to antibody-mediated toxicity, as no cytotoxicity was observed when tumor cells were treated with the meP antibodies alone (Supplementary Fig. S6C and S6D).

To explore the utility of meP antibodies to redirect meCAR T-cell specificity in vivo, NSG mice were engrafted intraperitoneally with HER2-negative CD20-engineered MDA468 (MDA468-CD20) tumors and treated with HER2-meCAR T cells and meP–anti-CD20-IgG every 2 days for 10 doses (Fig. 7A). Mice treated with the combination of meP–anti-CD20-IgG and HER2-meCAR T cells exhibited greater tumor reduction (Fig. 7B) and extended survival (Fig. 7C) when compared with mice treated with meCAR T cells alone or combined with CD20 antibody lacking meP (rituximab). The efficacy observed with anti-CD20 alone was unlikely due to redirected CAR T-cell activation, as meCAR T cells without antibody treatment showed no antitumor activity (Fig. 7B and C). Consistent with this, we did not observe activation of meCAR T cells upon CD20 antibody treatment (Supplementary Fig. S7A). We hypothesize that this effect may arise from transient interactions between the CD20 IgG and endogenous immune cells, such as myeloid-derived cells. However, the superior survival provided by meP–anti-CD20-IgG underscores its enhanced, targeted therapeutic potential. This observation led us to further optimize the system by utilizing an Fab format to minimize these IgG-mediated background effects.

Figure 7.

Dual targeting of HER2-meCAR T cells mediated by meP-αCD20 in a heterogeneous tumor model. A and B, Schematic of redirected targeting of HER2-meCAR T cells in vivo (A). NSG mice bearing ffLuc⁺ MDA468 tumors engineered to express CD20 (MDA468-CD20) were treated with HER2-meCAR CAR T cells, with or without meP-αCD20 (meP-IgG-H variant), followed by biophotonic imaging for tumor size over time. Mice treated with UTD T cells were used as the negative control (B). C, Kaplan–Meier survival analysis of each group. **, P < 0.01 (using a Mantel–Cox test). D and E, Schematic of dual targeting of HER2-meCAR T cells in vivo. ffLuc⁺ MDA468 tumor cells engineered to express CD20 or HER2 cells (D) were grafted to mimic tumoral heterogeneity (E) followed by HER2-meCAR T cells with or without αCD20 reagent (control αCD20 IgG, meP–αCD20-IgG, or meP–αCD20 Fab). F, Tumor growth was monitored by biophotonic imaging. G, Mean ± SD tumor flux values for each group. ***, P < 0.001 (using a two-way ANOVA test). H, Kaplan–Meier survival analysis of each group. ***, P < 0.001 (using a Mantel–Cox test). I and J, Blood samples collected on day 14 were evaluated for total T cells (I) and HER2-meCAR T cells (J) by flow cytometry. MFI, mean fluorescence intensity; RO, retro-orbital.

The Fc region of IgG supports serum stabilization and extended in vivo half-life, which is critical for therapeutic efficacy in vivo. However, the small size of Fab reagents offers advantages, including better tissue penetration and reduced steric hindrance. As our pilot studies using prototype meP–anti-CD20 Fab showed a lack of therapeutic efficacy (Supplementary Fig. S7B), we used MSA peptide (MSAP) to stabilize and increase in vivo half-life. We determined that fusing MSAP to the C-terminus of either the light chain (LCC) or heavy chain (HCC) of the Fab did not interfere with the meP–anti-CD20 Fab–redirected killing or IFNγ production by HER2-meCAR T cells (Supplementary Fig. S7C and S7D). Mice engrafted with MDA468-CD20 and treated with MSAP-fused meP–anti-CD20 Fab and HER2-meCAR T cells exhibited tumor reduction (Supplementary Fig. S7E and S7F). In this model, the LCC variant demonstrated better in vivo efficacy than the HCC variant (Supplementary Fig. S7E and S7F), and we used the meP–anti-CD20 Fab LCC format in subsequent experiments. Together, these studies highlight the feasibility of the meCAR platform to redirect T-cell specificity in vivo using meP-fused antibodies and underscore the importance of optimizing platform features, such as meP fusion position and antibody half-life, for maximal therapeutic efficacy.

We next evaluated the meCAR platform as a strategy to overcome tumor heterogeneity by developing a mixed-antigen solid tumor model expressing both CD20 and HER2 (Fig. 7D). For these studies, we engineered GFP/ffLuc⁺ MDA468 cells to express either HER2 (MDA468-HER2) or CD20 (MDA468-CD20) and intraperitoneally engrafted the tumor cells at a 1:4 ratio (MDA468-HER2:MDA468-CD20) in NSG mice, which yielded tumors with mixed antigen expression (Fig. 7E). Upon treatment with HER2-meCAR T cells alone or with control anti-CD20, we observed partial responses (Fig. 7F–H), with significant reduction of tumor burden and improved survival only seen upon treatment with HER2-meCAR T cells combined with meP–anti-CD20-IgG or meP–anti-CD20 Fab (Fig. 7F–H). Examination of circulating T cells in the blood at day 14 revealed comparable levels of HER2-meCAR T cells in all groups (Fig. 7I and J). Together, these results illustrate the dual-targeting flexibility of meCAR T cells combined with meP-fused Fab, both in terms of antigens and tumor models.

In silico evaluation of meP immunogenicity potential

Meditope, a non-human 12-mer cyclic peptide (CQFDLSTRRLKC), may have the potential to be immunogenic and elicit host immune responses in certain clinical settings. Although the immunogenicity of the meP cannot be fully assessed here in these preclinical studies and will be evaluated in future clinical trials, we assessed its potential immunogenicity using the in silico Immune Epitope Database (IEDB) Class I Immunogenicity tool to estimate peptide binding potential to human leukocyte antigen (HLA) class I and recognition by CD8 T cells (32). Negative immunogenicity scores indicate a low predicted likelihood of eliciting immune responses for the evaluated HLA alleles. As positive controls, the highly immunogenic cytomegalovirus pp65 peptide and the predicted immunogenic RR zipCAR peptide both yielded immunogenicity scores >0.5 across all six common HLA alleles examined. In contrast, the meP exhibited negative immunogenicity scores, similar to the FOS zipCAR peptide, which is also predicted to have low immunogenicity potential (Supplementary Fig. S8A and S8B; ref. 33). These results suggest that the predicted capacity of the meP to induce a robust T cell–mediated immune response is low under the current model parameters.

Discussion

CAR T-cell therapies have demonstrated transformative clinical activity, particularly in hematologic malignancies (2, 34, 35). However, several challenges continue to limit their broader applicability, including fixed antigen specificity, limited ability to regulate cell activity after adoptive transfer, and insufficient persistence and functionality in the immunosuppressive TME (36, 37). These limitations are especially pronounced in solid tumors but also affect CAR T-cell potency and safety across cancer types. To overcome these barriers, we developed a meditope-based, plug-and-play CAR T-cell platform that enabled precise and modular functional enhancement without altering the underlying CAR specificity. Our proof-of-concept studies demonstrated that meditope binding could be readily incorporated into Fab-based CAR constructs of diverse specificities and leveraged to introduce programmable features, such as selective expansion of CAR-engineered T cells both in vitro and in vivo using a meP-fused cytokine and tunable dual-antigen targeting via meP-fused antibodies. The meCAR/meP platform is not limited to CAR antigen specificity, tumor type, or meP conjugate and, therefore, may be broadly and rapidly applied across the CAR T-cell field.

The extent of CAR T-cell expansion and persistence following infusion strongly affects therapeutic efficacy. One strategy to enhance these properties involves co-expression of γ chain cytokines, such as IL2 and IL15, resulting in so-called “armored” CARs (38, 39). A recent clinical trial in various solid tumors demonstrated that armored GPC3 CAR T cells co-expressing IL15 exhibited enhanced cell expansion and disease control despite increased incidence of manageable cytokine release syndrome (40). However, these cytokine-armored approaches hardwire cytokine production into the CAR design, limiting the ability to modulate activity and raising concerns about altered phenotypes during manufacturing or increased toxicity after infusion, highlighting the importance of balancing safety and therapeutic potency. To address this limitation of armored CAR T cells, here we demonstrate that meP-IL15s, a soluble cytokine fused to the meP, selectively enhanced expansion and persistence of meCAR T cells in a dose-dependent manner, offering a modular and titratable system to support CAR T-cell function and improve antitumor efficacy. Our approach bypassed the need for additional transgenes to be expressed in the CAR cassette, simplifying manufacturing and reducing the risk of uncontrolled antigen-independent cytokine-driven expansion. Mechanistically, meP-IL15s selectively activated STAT5 signaling and proliferation in meCAR⁺ T cells through cis engagement, confirming the specificity of the interaction. Moreover, this external control system enabled dynamic regulation of CAR T-cell activity after infusion, a feature that could be tailored to patient needs or disease context and allows dosing to be optimized to reduce the risk of patient toxicities. The meP-based delivery system may be broadly adapted to deliver a range of cytokines or modulators, extending its potential as a flexible tool to optimize CAR T-cell performance.

Tumor heterogeneity remains a major challenge in CAR T-cell therapy, particularly in the context of solid tumors. Although single antigen–targeted CARs have shown remarkable clinical success in specific B-cell malignancies (1–3, 5), multiantigen targeting approaches are likely required to address tumor heterogeneity. Several groups have developed bispecific or tandem CARs with dual-antigen specificity, which have shown promising preclinical and clinical activity across several tumor types (41–49). However, tandem or bispecific CARs require permanent genetic incorporation of multitargeted CARs and extensive receptor optimization. In contrast, the meCAR platform decouples multiantigen specificity from CAR design, enabling programmable and switchable multiantigen targeting through externally administered meditope-fused antibodies. We show proof of concept that meP-fused antibody variants, including IgG, Fab, and nanobody, could effectively redirect meCAR T cells against alternative tumor antigens in vitro and in vivo while preserving engagement of the original target. As with other adapter-based systems, and consistent with the nonlinear concentration curves observed in our titration studies, excess meP antibodies may produce a hook effect that limits productive immune synapse formation, highlighting the requirement for careful dose optimization. Indeed, controlled dosing, pharmacokinetic modeling, and potentially tunable half-life engineering of meP antibodies will be important to maximize the efficacy of this platform while minimizing functional saturation. Overall, this plug-and-play meCAR system provides a versatile solution to tumor heterogeneity, allowing multiantigen targeting without the need for re-engineering or re-manufacturing of CAR T cells.

Similar to the meCAR platform, universal immune receptors (UIR) that separate antigen recognition from T-cell signaling such as Fabrack, UniCARs, switchable CAR (sCars), and Supra CARs, also provide flexibility in antigen targeting, enabling modification of the targeting strategy during the course of treatment to address tumor heterogeneity based on the dynamics of tumor antigen expression (33, 50–55). Additionally, both meCAR and UIR adapter systems have a built-in safety switch to reduce immune-related adverse events and on-target off-tumor risks, as adapter-directed CAR activity is dependent on the presence and dose of the biologic. However, meCARs are differentiated from UIRs as they also allow for integrated adapters for conferring additional functionality, including metabolic support and CAR T-cell tracking. By utilizing a highly specific cyclic meditope that docks into a unique engineered pocket of the meCAR-Fab, this platform avoids the tag-based designs of UIRs. As shown here, meCARs enable simultaneous cis presentation of cytokines (e.g., meP-IL15) for selective expansion, real-time fluorophore tracking (meP-647) for in situ monitoring of tumor–T-cell interactions, and redirection of meCAR T cells against alternative tumor antigens in vitro and in vivo while preserving engagement of the original target. Thus, this “soft-wired” CAR approach combines the advantages of standard “hard-wired” CARs for redirected antitumor activity while integrating the flexibility of UIRs for additional programmability, offering a robust solution for overcoming the barriers associated with treating solid tumors.

In summary, our meCAR platform leverages the meditope technology to confer versatile functionality to CAR T cells, offering new opportunities for imaging, manufacturing, in vivo expansion, and multiantigen targeting. Although the studies presented here serve as proof of concept, our findings support the broader applicability of this approach to CARs targeting diverse antigens and to a wide array meP-fused biologics. This platform provides a flexible framework to address several key limitations of current CAR T-cell therapies, particularly in the context of solid tumors. Looking forward, we anticipate that the meCAR platform can be extended to incorporate additional functionalities, including modifiers of the TME to overcome immunosuppression, safety switches to enable enhanced control, and labeled conjugates to facilitate dynamic imaging of CAR T cells.

Supplementary Material

Figure S1

Supplementary Fig. S1| Amino acid substitution for meCAR design and affinity characterization of meCAR variants.

Figure S2

Supplementary Fig. S2| in vivo efficacy of HER2 meCAR T cells in an OVCAR-3 xenograft model.

Figure S3

Supplementary Fig. S3| Design and characterization of meditope-enabled CARs targeting additional antigens.

Figure S4

Supplementary Fig.S4| meP-IL15s-mediated enrichment of meCAR T cells and assessment of toxicity.

Figure S5

Supplementary Fig. S5| Binding of meP-αCD20 on tumor triggers dual-targeting of HER2 meCAR T cells.

Figure S6

Supplementary Fig. S6| Dual-targeting of HER2-meCAR T cells by meP-αPD-L1 and meP-αCD20.

Figure S7

Supplementary Fig. S7| Optimized meP-αCD20 Fab shows improved antitumor activity.

Data Availability

The data generated and analyzed in this study are described in this article and its Supplementary Figures. The raw data in this study are available upon request from the corresponding authors.

Authors’ Disclosures

J. King reports current employment with Amgen. L.A. Stern reports grants from NIH and The Marcus Foundation during the conduct of the study. J.C. Williams reports other support from COH during the conduct of the study; other support from Xilio Therapeutics and Meditope Biosciences outside the submitted work; and a patent for EP 3 294 764 B1 issued to COH and a patent for other technologies not related to this work licensed out to SlamBio. C.E. Brown reports grants from The Marcus Foundation and NIH during the conduct of the study; personal fees from Dispatch Biotherapeutics and Aleta Biotherapeutics and other support from Mustang Bio outside the submitted work; and a patent for US12590137B2 issued. No disclosures were reported by the other authors.

Authors’ Contributions

C.-F. Kuo: Conceptualization, resources, data curation, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. Z. Tong: Data curation, formal analysis, investigation. Y.-C. Kuo: Conceptualization, resources. M. Park: Conceptualization, resources. J. King: Conceptualization, resources. K. Ly: Resources. B. Parcutela: Resources. L.A. Stern: Conceptualization, resources. Z. Wang: Resources. J.C. Hibbard: Resources. B. Aguilar: Resources. R. Starr: Resources. W.-C. Chang: Resources. J.R. Ostberg: Resources, writing–review and editing. D. Rossi: Resources, writing–review and editing. M.C. Clark: Resources, writing–review and editing. D. Alizadeh: Conceptualization, resources. S.J. Forman: Conceptualization, resources, supervision, funding acquisition. J.C. Williams: Conceptualization, resources, data curation, formal analysis, funding acquisition, writing–original draft, writing–review and editing. C.E. Brown: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, writing–original draft, writing–review and editing.