Cell culture

Buffy coats from de-identified healthy donor’s blood were purchased from the Austrian Red Cross, Vienna, Austria. Primary human T cells were isolated by negative selection using the RosetteSep Human T-Cell Enrichment Cocktail (STEMCELL Technologies) and cryopreserved in RPMI-1640 GlutaMAX medium (Thermo Scientific) supplemented with 20% FCS (Sigma-Aldrich) and 10% DMSO (Sigma-Aldrich) until further use. After thawing, primary human T cells were immediately activated with Dynabeads Human T-Activator αCD3/αCD28 beads (Thermo Scientific) at a 1:1 ratio according to the manufacturer’s instructions. T cells were expanded before the experiments for at least 14 days in T-cell medium, consisting of RPMI-1640 GlutaMAX supplemented with 10% FCS, 1% penicillin–streptomycin (Thermo Scientific) and 200 U/mL recombinant human IL-2 (Peprotech). Half of the medium was exchanged every other day and cell densities were kept between 0.4 and 2 × 106 cells/mL. Nalm-6 cells and Jurkat cells (gifts from Dr. Sabine Strehl and Dr. Michael Dworzak, respectively; CCRI, Vienna, Austria) were maintained in RPMI-1640 GlutaMAX supplemented with 10% FCS and 1% penicillin–streptomycin. Lenti-X 293T cells (Takara) were maintained in DMEM (Thermo Scientific) supplemented with 10% FCS. Nalm-6 target cell lines ± hEGFRt and/or ± hHER2t expression were established by transduction with a lentiviral vector encoding firefly luciferase (ffLuc), puromycin N-acetyl transferase and hEGFRt and/or hHER2t. A stable clone of Nalm-6 cells expressing both transgenes was established by limiting dilution cloning and subsequently was co-transduced with a lentiviral vector encoding ffLuc and enhanced GFP (eGFP) to ensure high expression of the luciferase reporter gene. Cell lines were regularly tested for mycoplasma contamination using the MycoAlert PLUS Mycoplasma Detection Kit (Lonza) and for squirrel monkey retrovirus contamination by PCR (performed by Labdia Labordiagnostik GmbH, Austria). Cell line authentication was performed by SNP-profiling at Multiplexion GmbH, Germany.

The homo- and heterodimerization of proteins in vitro was induced by the addition of AP20187 or AP21967 (both Takara) to the cell suspensions at a final concentration of 10 and 500 nM, respectively. The effective concentrations of both compounds were determined by the addition of increasing concentrations to Jurkat cells expressing hEGFRt, which could be homo- or heterodimerized (Supplementary Fig. 4d).

In vitro transcription and electroporation of mRNA

DNA encoding the respective transgene was amplified by PCR and subsequently transcribed in vitro with the mMessage mMachine T7 Ultra Kit (Ambion) according to the manufacturer’s instructions. The resulting mRNA was purified using the RNeasy Kit (Qiagen). Electroporation was performed with <1 × 107 cells/100 µl Opti-MEM (without phenol red; Thermo Scientific) using a Square wave protocol (1 pulse, 500 V) on the Gene Pulser Xcell Electroporation System (Biorad) and 4 mm electroporation cuvettes (VWR). The length of the pulse was 5 ms for primary human T cells and 3 ms for Jurkat cells. Primary human T cells were electroporated with 5 µg of CAR-mRNA and Jurkat cells were electroporated with different mRNA amounts as indicated. After electroporation, the cells were immediately recovered in warm cell growth medium and incubated over night at 37 °C.

Lentiviral transduction

To generate VSV-G pseudotyped lentivirus, Lenti-X 293T cells (Takara) were co-transfected with a third-generation puromycin-selectable pCDH expression vector (System Biosciences) and second-generation viral packaging plasmids pMD2.G and psPAX2 (Addgene plasmids #12259 and #12260, respectively; gifts from Didier Trono) using the PureFection Transfection Reagent (System Biosciences) according to the manufacturer’s instructions. Supernatants were collected on day 2 and 3 after transfection and were concentrated 100-fold using the Lenti-X Concentrator (Takara) according to the instructions provided by the manufacturer. Viral suspensions were resuspended in RPMI-1640 medium supplemented with 10% FCS and 1% penicillin–streptomycin and frozen at −80 °C. Twenty-four hours after their activation with αCD3/αCD28 beads (Thermo Scientific), primary human T cells were transduced in cell culture plates, which were coated according to the manufacturer’s instructions with RetroNectin (Takara). Thawed viral supernatant was added to the T cells (0.5 × 106 cells/mL) at a final dilution of 1:2. To ensure high and uniform expression of the transgenes, 1 µg/mL puromycin (Sigma-Aldrich) was added 3 days later and removed after 2 more days. Transduced T cells were expanded in AIMV medium (Thermo Scientific) supplemented with 2% Octaplas (Octapharma), 1% l-Glutamine (Thermo Scientific), 2.5% HEPES (Thermo Scientific), and 200 U/mL recombinant human IL-2. Half of the medium was exchanged every 2–3 days. Cell lines were transduced by exposure to varying concentrations of lentiviral supernatants for 3 days, followed by puromycin selection for 2 days with varying concentrations of puromycin and expansion in the respective growth medium.

Flow cytometric analysis

Cells were counted using AccuCheck Counting Beads (Thermo Scientific). Propidium iodide (Sigma-Aldrich) was added for exclusion of dead cells. Unspecific antibody binding was prevented by preincubation of the cells for 10 min at 4 °C in FACS buffer [i.e., PBS (Thermo Scientific), 0.2% human albumin (CSL Behring) and 0.02% sodium azide (Merck KGaA)] supplemented with 10% (v/v) human serum. Alternatively, for subsequent detection of Fc-containing CARs, preincubation was done in antihuman FACS buffer (i.e., PBS, 10% FCS and 0.02% sodium azide) supplemented with 10% (v/v) murine IgG1-κ (clone MOPC21; 1 mg/mL; Sigma-Aldrich). This preincubation was omitted for the detection of the FMC63-derived αCD19-CAR by the protein L-biotin conjugate (GenScript; dilution 1:167). The following antibodies were used: αFLAG tag (PE; clone L5; BioLegend; dilution 1:167), αFLAG tag (APC; clone L5; BioLegend; dilution 1:167), αStrep II tag (biotin; clone 5A9F9; GenScript; dilution 1:500), αhEGFR (APC and PE; clone AY13; BioLegend; dilution 1:50), and αhHER2 (PE; clone 24D2; BioLegend; dilution 1:50), αhCD19 (BV421; clone HIB19, BioLegend; dilution 1:167), αhCD45 (PerCP; clone 2D1; BD Biosciences; dilution 1:50), αhCD3 (PE-Cy7; clone SK7; BD Biosciences; dilution 1:50), αGFP (PE; clone FM264G; BioLegend; dilution 1:50). Binding of hEGFR-Fc and mEGFR-Fc (both R&D) to hEGFR-specific CARs was detected by an antibody directed against human IgG1-Fc (PE; clone JDC-10; Southern Biotech; dilution 1:25). Cells were incubated with the respective antibodies or protein L for 25 min at 4 °C and then washed twice with ice-cold FACS buffer or—if Fc-containing CARs were detected—antihuman FACS buffer. Biotinylated antibodies and protein L were further incubated with streptavidin APC or streptavidin PE (both Thermo Scientific; dilution 1:250) for 25 min at 4 °C and then again washed twice. For the analysis of single-cell suspensions of mouse organs, dead cells were excluded by staining with the Fixable Viability Dye eFluor™ 780 (Thermo Scientific). Subsequently, the cells were acquired on an LSR Fortessa instrument (BD Biosciences) and analyzed using the FlowJo Software (Version 10.6.1; FlowJo, LLC) and the BD FACSDivaTM Software V8.0.1. Nontransfected cells or isotype antibodies served as negative controls.

Quantitation of target antigen density

The surface density of target antigen molecules on Jurkat cells was quantified using the QuantiBRITE Phycoerythrin (PE) Fluorescence Quantitation Kit (Becton Dickinson) according to the manufacturer´s instructions. Briefly, the cells were stained with a saturating concentration of a PE-labeled αhEGFR-antibody (clone AY13; BioLegend). Subsequently, the geometric mean of the fluorescence intensity was determined, subjected to background subtraction using unstained or isotype-labeled cells and then used to estimate the number of antibodies bound per cell (ABC). ABC values were corrected for the PE-conjugation efficiency of the respective antibody (i.e., PE dye-to-antibody ratio) yielding effective surface densities.

Generation of mutant binders by alanine scanning

Site-directed mutagenesis of the hEGFR- and hHER2-specific binder scaffolds was performed using the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent Genomics), according to the manufacturer’s instructions. Mutagenic primers were designed using the QuikChange Primer Design Software (Agilent Genomics) and synthesized by Biomers, Germany. Primer sequences can be found in Supplementary Data 3. In the case of the hHER2-specific binder, the affibody zHER2-AK was generated from the parental affibody zHER2:4, which binds to hHER2 with an affinity of 50 nM45, prior to the alanine scan by introducing mutations N23A and S33K for preventing N-glycosylation66.

Determination of binding affinities on cell membranes

Jurkat cells with high level expression of hEGFRt were obtained by electroporation with 3 µg of hEGFRt-mRNA. On the next day, the cells were washed with PBS, resuspended in PBSA [PBS supplemented with 0.1% BSA (Sigma-Aldrich)] and incubated in 96-well plates for 1 h at 4 °C with varying concentrations of fluorescent chimeric hEGFR-binder proteins (i.e., E11.4.1 or mutants thereof fused to sfGFP). The plates were then centrifuged (450 × g, 7 min, 4 °C), the supernatant discarded and the cells resuspended in ice-cold PBSA immediately prior to acquisition on an LSR Fortessa instrument. The cells were kept on ice during the analysis to avoid endocytosis. Kd values were calculated by curve fitting using Microsoft Excel (Microsoft Corporation).

Construction of transgenes

A detailed description of all used transgenes is given in Supplementary Data 1. hEGFRt and hHER2t were generated using the Addgene plasmids #1101167 and #1625768, respectively (gifts from Matthew Meyerson and Mien-Chie Hung). sfGFP was derived from the Addgene plasmid #5473769 (gift from Michael Davidson & Geoffrey Waldo). Cytoplasmic domains of CD28, ICOS, OX40, and CD2 were derived from cDNA clones (Sino Biological). The cytoplasmic domain of CD3ζ contained a Q65K mutation (Uniprot P20963-3). Synthetic DNAs were synthesized by GeneArt (Thermo Scientific). Assembly of DNA molecules into complete constructs was performed using the Gibson Assembly Master Mix (New England BioLabs) according to the manufacturer’s instructions.

Recombinant expression of binder proteins

To determine their affinities, the binders were expressed recombinantly as soluble proteins with an N-terminal hexahistidine and a C-terminal sfGFP fusion tag using the pE-SUMO vector (Life Sensors), in which the SUMO-tag was deleted. Briefly, Escherichia coli cells (Tuner DE3) were transformed with respective plasmids using heat shock transformation. After overnight incubation in lysogeny broth (LB) medium at 37 °C, cultures were diluted 1:100 in terrific broth medium supplemented with kanamycin (50 µg/mL) and were kept shaking at 37 °C. Expression of the transgene was started when cultures reached an A600 of 2 by addition of 1 mM of isopropyl β-D-1-thiogalactopyranoside (IPTG) followed by overnight incubation at 20 °C. Transformed cells were harvested by centrifugation (5,000 × g, 20 min, 4 °C) and resuspended in sonication buffer (50 mM sodium phosphate, 300 mM NaCl, 3% glycerol, 1% Triton X-100, pH 8.0). Cell disruption was accomplished by sonication (2 × 90 s, duty cycle 50%, amplitude set to 5) followed by removal of cell debris by centrifugation (20,000 × g, 30 min, 4 °C). Hexahistidine-tagged proteins were extracted from crude cell lysates by immobilized metal affinity chromatography using TALON metal affinity resin (Takara). Sonicated supernatants were supplemented with 10 mM imidazole and applied onto the resin twice, followed by repeated washing steps with equilibration buffer (50 mM sodium phosphate, 300 mM NaCl, pH 8.0) supplemented with increasing concentrations of imidazole (5–15 mM). Purified proteins were eluted by washing the column with equilibration buffer with 250 mM imidazole. Buffer exchange to PBS was performed with Amicon Ultra-15 10 K centrifugal filters (Merck Millipore). Affibody-based binder scaffolds were further purified by performing a preparative SEC using a Superdex 200 column (10 mm × 300 mm, GE Healthcare). Purified proteins were directly frozen at −80 °C. Truncated human VEGF (Uniprot P15692 amino acids 40–134) was expressed as previously described46. Briefly, E. coli BL21 (DE3) cells were transformed with the pJ414 vector (ATUM) including the gene encoding for truncated VEGF. LB medium supplemented with ampicillin (100 µg/mL) was inoculated with an overnight culture. Once the culture reached an A600 of 2 at 37 °C, the overexpression was induced by addition of 1 mM IPTG. After 4 h of shaking at 37 °C, cells were harvested by centrifugation (5,000 × g, 20 min, 4 °C). The pellet was resuspended in 20 mM TRIS-HCl buffer (pH 7.5) including 5 mM ethylenediaminetetraacetic acid (EDTA), incubated for 10 min and disrupted by ultrasonication with a Vibra-Cell 375 ultrasonic processor (Sonics & Materials, Inc.). After two washing steps using the same buffer, the pellet was resuspended in 20 mL unfolding buffer (20 mM TRIS-HCl, 5 mM EDTA, 7.5 M urea, 4 mM dithiothreitol, pH 7.5), stirred for 2 h at room temperature followed by centrifugation (39,000 × g, 25 min, 4 °C). The supernatant containing the unfolded protein was diluted tenfold into refolding buffer (20 mM TRIS-HCl, 7 mM CuCl2, pH 8.4) and stirred overnight at room temperature. Following a dialysis step against 20 mM TRIS-HCl (pH 8.0), the protein was purified using a 6 mL Resource Q column, a 1 mL HiTrap Phenyl FF (Low Sub) column and a HiLoad 16/600 Superdex 200 pg (all from GE Healthcare) according to the manufacturer’s instructions. Pure VEGF samples were stored at −80 °C.

Size exclusion chromatography

For analytical SEC analysis, recombinant proteins were diluted in SEC running buffer (PBS supplemented with 200 mM NaCl) and filtered through a 0.1 µm Ultrafree MC VV centrifugal filter (Merck Millipore). Subsequently, 25 µg protein was loaded onto a Superdex 200 column (10 mm × 300 mm, GE Healthcare) connected to an HPLC Prominence LC20 System (Shimadzu) at a flow rate of 0.75 mL/min at 25 °C.

Surface plasmon resonance

SPR analysis was performed with a Biacore T200 instrument (GE Healthcare). All experiments were performed in degassed and filtered PBS, pH 7.4, supplemented with 0.1% BSA (Sigma-Aldrich) and 0.05% Tween-20 (Merck Millipore) at 25 °C. Immobilization of hEGFR-Fc and hHER2-Fc (both R&D) was performed on a Protein A sensor chip (GE Healthcare) at a flow rate of 10 µL/min for 60 s at a concentration of 6.67 and 4 µg/mL, respectively, yielding a density of ~400 response units. Five different concentrations of the respective binder scaffold (depending on the expected Kd value) were injected at a flow rate of 30 µL/min for 15 s (for binders E11.4.1, E11.4.1-G25A, E11.4.1-G32A, and zHER2-AK-R10A) or 60 s (for zHER2-AK) in the single-cycle kinetic mode. Subsequently, a dissociation step was performed for 30 s (all E11.4.1-based binders), 60 s (zHER2-AK-R10A), or 180 s (zHER2-AK). Protein A sensor chips were regenerated by applying 10 mM Glycine-HCl, pH 1.5, at a flow rate of 30 µL/min for 30 s. The Kd value was either obtained by steady state binding analysis or by fitting the sensorgram to a 1:1 Langmuir model in the kinetic mode using the Biacore T200 Evaluation Software (GE Healthcare).

Western blotting and crosslinking of membrane proteins

Crosslinking of membrane proteins was performed with disuccinimidyl suberate (DSS) (Thermo Scientific) according to the manufacturer’s instructions. Briefly, Jurkat cells were washed three times with ice-cold PBS. Cells were resuspended in PBS to a cell concentration of 20 × 106 cells/mL. Subsequently, DSS was added at a concentration of 5 mM and the reaction was incubated for 30 min at room temperature. Chemical crosslinking was stopped by addition of the Quench Solution (1 M TRIS, pH 7.5), yielding a final concentration of 20 mM TRIS. Cells were washed once in PBS and lysed in ice-cold RIPA buffer (Sigma-Aldrich), supplemented with protease inhibitors (Halt Protease Inhibitor Cocktail; Thermo Scientific) and nucleases (Pierce Universal Nuclease; Thermo Scientific). The lysate was incubated on ice for 5 min. Subsequently, samples were denatured for 5 min at 50 °C in loading buffer (LDS Sample Buffer; Thermo Scientific) and frozen at −20 °C. Cleared lysates (14,000 × g, 10 min, 4 °C) were resolved on 4–20% gradient SDS-PAGE gels (Mini-PROTEAN TGX Precast Protein Gels; BioRad) under nonreducing conditions. Separated proteins were transferred to 0.45 µm nitrocellulose membrane (GE Healthcare) and blocking was performed with Western Blocking Reagent (Roche) for 1 h at room temperature. Membranes were probed with an antibody against hEGFR (clone 528; 1:500 dilution; Thermo Scientific) and incubated overnight at 4 °C. Subsequently, membranes were incubated with a secondary goat antimouse horseradish peroxidase-conjugated antibody (Thermo Scientific) for 1 h at room temperature and bands were developed using the SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific), according to the manufacturer’s instructions. Loading control was performed by probing the membrane with an antibody against GAPDH (clone ab9485; 1:5,000; Abcam) for 1 h at room temperature. Detection was performed by incubating the membrane with a secondary goat antirabbit DyLight 800-conjugated antibody (Thermo Scientific) for 1 h at room temperature and visualizing the bands on the Odyssey Imaging System with the Odyssey Infrared Imaging System software (Version 3.0, LI-COR).

Cytotoxicity assay

Cytotoxicity of primary human T cells was assayed with either a luciferase- or a flow cytometry-based assay. Luciferase-based assay: 10,000 target cells (i.e., luciferase-expressing Nalm-6 or Jurkat cells) were co-cultured for 4 h at 37 °C with 20,000 primary human T cells (= E:T cell ratio 2:1, unless stated otherwise) in white round-bottom 96-well plates (Sigma-Aldrich) in 100 µL RPMI medium without phenol red (Thermo Scientific) supplemented with 10% FCS and 1% penicillin–streptomycin. After co-culture, viable target cells were quantified by determining the luciferase activity. Hereto, the culture plates were equilibrated for 10 min to room temperature and subsequently luciferin (Xenolight D-luciferin; Perkin Elmer) was added to the cell suspension (150 µg/mL final concentration). Bioluminescence was quantified after 20 min of incubation at room temperature on an ENSPIRE Multimode plate reader (Perkin Elmer). Specific lysis was calculated with the following formula:

$$% ,{mathrm{specific}},{mathrm{lysis}} = 100 – left( {frac{{frac{{{mathrm{RLU}},{mathrm{antigen}}^{{mathrm{pos}}}{mathrm{target}} , + , {mathrm{effector}},{mathrm{cells}}}}{{{mathrm{RLU}},{mathrm{antigen}}^{{mathrm{pos}}}{mathrm{target}},{mathrm{cells}},{mathrm{only}}}}}}{{left( {frac{{{mathrm{RLU}},{mathrm{antigen}}^{{mathrm{neg}}}{mathrm{target}} , + , {mathrm{effector}},{mathrm{cells}}}}{{{mathrm{RLU}},{mathrm{antigen}}^{{mathrm{neg}}}{mathrm{target}},{mathrm{cells}},{mathrm{only}}}}} right)}}} right) times 100.$$


Specific lysis in Fig. 5a, b was calculated with the following formula:

$${mathrm{% }},{mathrm{specific}},{mathrm{lysis}} = 100 – left( {frac{{{mathrm{RLU}},{mathrm{target}} + {mathrm{effector}},{mathrm{cells}}}}{{left( {{mathrm{RLU}},{mathrm{target}},{mathrm{cells}},{mathrm{only}}} right)}}} right) times 100.$$


Flow cytometry-based cytotoxicity assay: the assay principle is based on the determination of the ratio of antigenpos and antigenneg target cells that expressed either a green or red fluorescent protein, respectively, and were mixed 1:1 before co-culture with T cells. For this purpose, Jurkat cells were electroporated on the day before the assay (1) with mRNA encoding eGFP and mRNA encoding the respective target antigen, and alternatively (2) with mRNA encoding mCherry. Those cells were co-cultured for 4 h at 37 °C in round-bottom 96-well plates with 40,000 primary human T cells at an E:T cell ratio of 4:1:1 (i.e., 40,000 CAR-T cells plus 10,000 antigenpos and 10,000 antigenneg Jurkat cells per well). Target cells not co-cultured with CAR-T cells served as a negative control (targets only). After co-culture, the ratio of eGFP- to mCherry-expressing target cells was determined by flow cytometric analysis. Specific lysis was calculated with the following formula:

$$% ,{mathrm{specific}},{mathrm{lysis}} = 100 – left( {frac{{frac{{% ,{mathrm{eGFP}}^{{mathrm{pos}}},{mathrm{cells}},{mathrm{of}},{mathrm{the}},{mathrm{sample}}}}{{% ,{mathrm{mCherry}}^{{mathrm{pos}}},{mathrm{cells}},{mathrm{of}},{mathrm{the}},{mathrm{sample}}}}}}{{left( {frac{{% ,{mathrm{eGFP}}^{{mathrm{pos}}},{mathrm{cells}},{mathrm{of}}, {hbox{”}}{mathrm{targets}},{mathrm{only}}{hbox{”}} ,{mathrm{control}}}}{{% ,{mathrm{mCherry}}^{{mathrm{pos}}},{mathrm{cells}},{mathrm{of}}, {hbox{”}}{mathrm{targets}},{mathrm{only}}{hbox{”}} ,{mathrm{control}}}}} right)}}} right) times 100.$$


Cytokine secretion by CAR-T cells

Cytokine secretion of CAR-T cells was determined by co-cultivation with target cells at an E:T cell ratio of 2:1 in flat-bottom 96-well plates for 4 h at 37 °C. Alternatively, cytokine secretion was quantified in the supernatants from cytotoxicity experiments, described above. Supernatants were centrifuged (450 × g, 7 min, 4 °C) and stored at −80 °C. IFN-γ was quantified by enzyme-linked immunosorbent assay (ELISA) using the Human IFN-γ ELISA Ready-SET-Go! Kit (eBioscience/Thermo Scientific) according to the manufacturer’s instructions. Analysis was performed on the ENSPIRE Multimode plate reader (Perkin Elmer).

In vivo experiments

NOD.Cg-Prkdcscid Il2rgtm1WJI/SzJ (NSG, The Jackson Laboratory) mice were bred in the Anna Spiegel facility for animal breeding (Vienna, Austria) under standardized conditions (room temperature 22 ± 2 °C, humidity 55 ± 10%, air change rate of 15 and a dark/light cycle of 12 h). Subsequently, mice were transferred to the Preclinical Imaging Laboratory (PIL) or the Center for Biomedical Research of the Medical University of Vienna. All procedures were approved by the Magistratsabteilung 58, Vienna, Austria (GZ: 319093/2014/16) and the Federal Ministry Republic of Austria for Education, Science, and Research (BMBWF-66.009/0243-V/3b/2019). Primary human T cells were lentivirally transduced for CAR expression and expanded for 15–17 days to generate sufficient cell numbers. CAR-T cells directed against hEGFR showed no cross-reactivity with mEGFR as confirmed with recombinant mEGFR-Fc (R&D) (Supplementary Fig. 7a). Nalm-6 cells were engineered to express high levels of the hEGFRt, hHER2t, or both (intracellularly fused to the wild-type FKBP12 domain) and ffLuc [hEGFRt-Nalm-6 (SEQ-ID 20), hHER2t-Nalm-6 (SEQ-ID 23), and hHER2t-hEGFRt-Nalm-6 (SEQ-ID 20 and 23), respectively]. After filtering through a 35 µm cell strainer (Corning Falcon), 0.5 × 106 of those Nalm-6 cells (in 100 µL PBS) were injected i.v. into the tail veins of NSG mice (male and female, 8–20 weeks of age). Three or 5 days later (as indicated), when tumor burden became detectable by bioluminescence imaging (BLI), CAR-T cells (10 × 106 cells in 100 µL PBS) were injected i.v. into the tail veins of the NSG mice. Dimerization of the ON-switch AvidCAR was induced by intraperitoneal (i.p.) administration of 2 mg/kg AP20187 (Takara). AP20187 was prepared by dissolution in ethanol yielding a concentration of 12.5 mg/mL and further dilution to the final working concentration of 0.5 mg/mL in vehicle solution [4% ethanol, 10% PEG-400 and 1.7% Tween-80 (both Sigma-Aldrich) in water]. Those working stocks of AP20187 were sterile filtered and used for injection within 30 min. Injection of only the vehicle solution served as control. The mice were monitored daily and sacrificed by cervical dislocation at the first sign of disease (weight loss, rough fur, beginning paralysis).

In vivo bioluminescence imaging

BLI was performed at the PIL (Medical University of Vienna) using an IVIS Spectrum In Vivo Imaging System (Perkin Elmer). D-luciferin (Perkin Elmer) was freshly dissolved in PBS to a final concentration of 15 mg/mL and sterile filtered. Mice were anesthetized with isoflurane (Abbvie) and received i.p. injections of the luciferin solution (dosage of 150 mg/kg body weight). After 15–20 min, mice were transferred to the IVIS Imaging System and bioluminescence was measured immediately in medium binning mode with an automatic acquisition time (ranging from 1 s to 2 min) to obtain unsaturated images. The total photon flux was determined within the region of interest which encompassed the entire body of the mouse. Tumor growth was monitored every 3–4 days. Living Image Software (Caliper) was used to analyse the data.

In vivo model for the AND-gate AvidCAR

NSG mice were injected via tail vein with 0.5 × 106 cells of a 1:1:1:1 mixture of Nalm-6, hEGFRt-Nalm-6, hHER2t-Nalm-6 and hHER2t-hEGFRt-Nalm-6 cells and 3 days later with 10 × 106 CAR-T cells or mock-T cells, as indicated. Mice were sacrificed 13 days after tumor injection. Bone marrow single-cell suspensions were obtained by flushing one femur with ~10 mL sterile PBS and passing through a 70 µm cell strainer (Corning Falcon). After disruption of the liver through a 70 µm cell strainer and several washing steps, lymphocytes were separated from hepatocytes using a 33.75% Percoll gradient (GE Healthcare). Spleen single-cell suspensions were obtained by two consecutive passages through 70 µm cell strainers. Blood was drawn retro-orbitally from isoflurane-anesthesized mice and collected in EDTA-containing tubes (Greiner). Fifty µL blood was used for the staining with fluorescent antibodies, followed by red blood cell lysis using an RBC lysis/fixation solution (Biolegend).

Mathematical modeling

The multivalent interaction between a bivalent antigen and a bivalent CAR that can undergo kinetic proofreading leads to a large number of distinct chemical states. Therefore, it is not practical to manually enumerate them. To overcome this, we used the rule-based framework of BioNetGen70 to define five elementary reaction: intermolecular binding (kon), intermolecular unbinding (koff), intramolecular binding (σ kon), and the kinetic proofreading phosphorylation rate (kp). We have assumed that (1) the intramolecular unbinding rate (i.e., the unbinding rate between a single CAR and antigen when both CARs are bound to antigen) is equal to the intermolecular unbinding (i.e., unbinding rate of a single CAR and antigen when only a single CAR in the dimer is bound to antigen) and (2) the kinetic proofreading phosphorylation rate can only take place when at least one CAR in the dimer is bound to antigen. When a single CAR in the dimer is bound, the binding rate of the second CAR in the dimer to the second antigen in the dimer is expected to be dependent on the intermolecular on-rate (since the same binding interface is formed) multiplied by a factor that accounts for the local concentration (and potentially other factors, such as structural alignment) that we define to be σ with units of concentration. This parameter controls the avidity of the CAR with σ = 0 producing a monovalent CAR and increasing values of σ lead to increasing avidity. Collectively, these elementary reactions lead BioNetGen to 72 chemical states that were exported as a system of 72 coupled nonlinear coupled ordinary differential equations. The initial conditions were 0 for all chemical species except the concentration of the CAR (1,000/μm2) and antigen (varied as indicated). The parameter values used were kon = 5 × 10−4 μm/s, koff = 10/s, and kp = 0.1/s. These equations were solved numerically in Matlab (Mathworks, MA) to produce the panels in Supplementary Fig. 3b showing the bound receptor and phosphorylated receptor. To calculate the cellular response in Supplementary Fig. 3b, we have assumed a saturating Hill function connecting phosphorylated receptor to the cellular response using a threshold of (kthres = 2/μm2). Although the parameter values are expected to alter the absolute concentration of antigen inducing receptor binding, phosphorylation, and cellular response, they do not alter our main qualitative conclusion which is that increasing the cooperativity factor of dimeric binding (σ) increases the amount of phosphorylated receptor and the cellular response without increasing CAR binding. The bngl file defining the model along with the m-file used to solve it can be found in Supplementary Data 4 that also includes all parameter values.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 7 software for Windows (GraphPad Software Inc.) and SAS software (Version 9.4, SAS Institute). Graphs were generated using GraphPad Prism 7 software. Data are presented as individual data points or as means ± SD. Statistical analysis was performed using two-tailed paired t-tests, two-tailed ratio-paired t-test, or two-tailed unpaired t-test, as indicated. Statistical analysis of Figs. 4b and 6b was done with a mixed model to compare fixed-effect means taking into account that the data are correlated. Statistical analysis of Figs. 5d and 6d was done by log-rank analysis or two-way ANOVA, respectively.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.


Source link

Leave a Reply

Your email address will not be published. Required fields are marked *