Inefficient exploitation of accessory receptors reduces the sensitivity of chimeric antigen receptors

Significance T cells are engineered to recognize cancer antigens using chimeric antigen receptors (CARs). This therapy is approved to treat B cell cancers, but patients relapse with B cells that express low levels of the antigen. It is now clear that CARs have a profound defect in antigen sensitivity. They require > 100-fold more antigen to activate T cells compared to their native T cell antigen receptors (TCRs). Here we demonstrate that the antigen sensitivity defect of CARs is a result of their inability to fully exploit adhesion receptors. Modifying CARs so that they more closely resemble the TCR allows them to efficiently exploit adhesion receptors and fully restores their antigen sensitivity. The work suggests to improve CAR-T cell therapy.

T cells were discriminated from T2 cells by the absence of Tag-It Violet stain. Single T cells were identified on the basis of size and subsequent analysis performed on this population.
Solid-phase plate stimulation. Pierce Streptavidin Coated High Capacity 96 well plates (Thermo Fisher) were washed with PBS and dilutions of biotinylated pMHC in PBS were added to each well in a 50 volume and incubated for 90 minutes at room temperature. Subsequently, plates were washed again with PBS and biotinylated accessory molecules (CD58, ICAM-1, CD86, CD70, 41BBL) were added at a fixed dose of 250 in 50. Plates were again incubated for 90 minutes and then washed with PBS.
Downregulation with Brefeldin A. Cells were prepared as described in the section 'Plate Stimulation', but seeded in a smaller volume of 150 uL/well. Brefeldin A 1000X (BioLegend) was diluted to 4X in sufficient volume for the number of seeded wells and added 50 uL/well after 5 hours of incubation (final concentration 1X, 0.005 mg/ml). Plates were then incubated for the remaining 19 hours of the assay. Harvesting and staining is as in 'Plate Stimulation' with the difference that prior to antibody staining the samples were stained with Zombie nIR viability stain (BioLegend) 1:1000 in 50 uL/well of PBS, and subsequently washed with 200 uL of PBS 1% BSA before proceeding. Samples were not fixed and were run immediately on a flow cytometer. Dead cells were excluded from the subsequent analysis.
APC stimulation (co-culture with U87 cells). 25000 U87 cells were seeded in a tissue culture treated flat-bottom 96 well plate and grown overnight. On the following day the media was removed from these cells and they were incubated with peptides prepared to the appropriate concentration in complete DMEM (DMEM supplemented with 10% v/v FBS, 100 penicillin, 100 streptomycin) for 1 hour at 37.
If blocking antibodies were used then the appropriate amount of T cells were incubated for 30 minutes prior to addition to the U87 cells with either anti-IgG1 Isotype control (BioLegend, Clone MOPC-21), anti-CD58 (BioLegend, Clone TS2/9) or anti-ICAM1 (eBioscience, Clone HA58) at a concentration of 10. Alternatively, both anti-CD58 and anti-ICAM1 together at a concentration of 5 each (total antibody concentration 10).
Peptide containing media was then removed and 50,000 T cells per well were added. The co-culture was then spun for 2 minutes at 50, and incubated for 4 hours at 37. After this period a fraction of supernatant was removed for cytokine ELISAs and stored at -20. EDTA was added to the remaining supernatant (final concentration 2.5) and cells were detached by pipetting.
T cells were discriminated from U87 cells by CD45 staining and/or an assessment of size and complexity. Single T cells were identified on the basis of size and subsequent analysis performed on this population.
Flow cytometry. Tetramers were produced using refolded monomeric biotinylated pMHC and streptavidin-PE (Biolegend) at a 1:4 molar ratio. Streptavidin-PE was added in 10 steps with a 10 minute incubation at room temperature between each addition. 0.05-0.1% sodium azide was added for preservation and tetramers were kept for up to 3 months at 4.
Samples were analysed using a BD LSR Fortessa X-20 (BD Biosciences) or CytoFLEX LX (Beckman Coulter) flow cytometer and data analysis was performed using FlowJo v10 (BD Biosciences).
Electroporation of 868 TCR. 868 TCR alpha and beta chains were amplified using PCR, adding a T7 promoter at the 5 end. The resulting PCR product was 'cleaned up' using a NucleoSpin Gel and PCR clean-up kit (Macherey-Nagel). Capped and Poly(A) tailed mRNA was produced from this PCR product using a mMESSAGE mMACHINE™ T7 ULTRA Transcription Kit (ThermoFisher). mRNA was collected by lithium chloride precipitation, quality checked by gel electrophoresis and stored in single use aliquots at -80.
For electroporation, T cells are collected and washed 3x with Opti-MEM (Gibco) and resuspended at a concentration of 25e6. 5e6 with 2 per million cells of each of the RNA for the TCR, and chains. Cells were then aliquoted in 200 into an electroporation cuvette (Cuvette Plus 2mm gap BTX). Electroporation is performed using an ECM 830 Square Wave electroporation system (BTX) at 300V for 2. Cells are then transferred to pre-warmed complete RPMI at a density of 1e6. Electroporated cells are used in assays 24 hours later. Immobilisation Assay. Following a plate stimulation assay, after cells were collected, plates were washed 3 times with PBS 0.05% TWEEN 20 ('PBST') and then stained with anti-HLA-A,B,C (clone W6/32, dilution 1:1000) in PBS for 2 hours at room temperature. Plates were then washed 3x with PBST and stained with secondary goat anti-mouse IgG IRDye 800CW (LI-COR) in PBS for a further 2 hours. Finally plates were washed one more time with PBST and then imaged using a LICOR Odyssey Sa (LI-COR). Integrated intensity per well is reported.
ELISAs. Invitrogen Uncoated ELISA kits for IFN (Thermo Fisher) were used following the manufacturer's protocol. Supernatants were either used immediately for ELISAs post-harvesting or stored at -20 for up-to 2 weeks. Supernatants were diluted using an empirically determined ratio before use in an ELISA so that quantities of assessed cytokines fell within the linear range of the kits.
Surface Plasmon Resonance. D52N-pMHC interactions were analysed on a Biacore T200 instrument (GE Healthcare Life Sciences) at 37°C and a flow rate of 30 l/min. Running buffer was HBS-EP (10 mM HEPES pH 7.4, 150 mM NaCl, 3 mM EDTA, 0.005% v/v Tween-20). Streptavidin was coupled to CM5 sensor chips using an amino coupling kit (GE Healthcare Life Sciences) to near saturation, typically 10,000-12,000 response units (RU). Biotinylated pMHCs (47 kDa) were injected into the experimental flow cells (FCs) for different lengths of time to produce desired immobilisation levels (300-1000 RU). FC1 and FC3 were used as reference FCs for FC2 and FC4, respectively. Biotinylated ECD of CD58 (24 kDa + ∼25 kDa glycosylation) was immobilised in the reference FCs at levels matching those of pMHCs. Excess streptavidin was blocked with two 40 s (D52N STAR) or 60 (D52N scFv) injections of 250 M biotin (Avidity). Before injections of purified D52N, the chip surface was conditioned with eight injections of the running buffer. Dilution series of D52N were injected simultaneously in all FCs starting from the lowest concentration, which was injected again after the highest concentration to confirm stability of pMHC on the chip surface. The duration of injections (20 or 180 s) was the same for conditioning and D52N injections. After every 2 or 3 D52N injections, buffer was injected to generate data for double referencing. In addition to subtracting the signal from the reference FC (single referencing), all D52N binding data were double referenced versus the average of the closest buffer injections before and after D52N injection to correct for small differences in signal between flow cells. D52N binding versus D52N concentration was fitted with the following model: , where B is the response (binding) and Bmax is the maximal binding.
Sequences. D52N scFvs with the following sequence were produced by Absolute Antibody Ltd. D52N scFv: The D52N-IgG1 CARs contain a 'HNG spacer sequence' derived from the IgG1 hinge region, described in (4), and spliced with a spacer region from the CH2-CH3 regions of IgG1 as described in (5). HNG Spacer: DPAEPKSPDKTHTCPPCP The 1G4 TCR and chains are joined by a P2A linker peptide with an additional spacer and furin cleavage site, as described in (6). The sequence is given below.

Furin-P2A: GSRAKRSGSGATNFSLLKQAGDVEENPGP
Independent experiments and data analysis. To produce independent measurements of EC50 (individual data points in figure panels) for a given antigen receptor, we produced a new batch of lentivirus which was used to transduce T cells isolated from a new leukocyte cone, which is provided by the National Health Service (NHS) in the UK and is obtained from human blood donors.
In each independent experiment, we included the TCR and one or more CARs to be tested and used pMHC antigen tetramers to evaluate the percent of T cells expressing each antigen receptor (Fig. SS2C, transduction efficiency) and the surface level (gMFI of T cells expressing the antigen receptor) for each antigen receptor relative to the TCR (Fig. SS2D). Although we observed variations in the transduction efficiency, the surface level of each antigen receptor was always at the same level or higher compared to the TCR.
As a result of differences in the transduction efficiency, we observed differences in the maximum number of T cells that could upregulate CD69 or 4-1BB across independent experiments for the same antigen receptor or across different antigen receptors (see y-axis in Fig. 2B,C for example). These differences reflect the percent of T cells that express the antigen receptor and can therefore respond to the presented antigen. Importantly, our study focused on measuring antigen sensitivity (), which is defined as the concentration of antigen required to elicit half-maximal response. Therefore, variations in the maximum number of T cells able to respond are taken into account when measuring an .
Statistical analysis was performed using Prism (GraphPad Software) or Excel (Microsoft). Curve fitting was performed using the robust nonlinear regression function in Prism or MATLAB (MathWorks) and the EC50 extracted from the fitted curves. Data was excluded from analysis if the computed fit was reported as 'ambiguous' in Prism, if the fit did not converge in 1000 iterations, or if the computed EC50 was outside of the tested ligand concentration.
All data is included in the manuscript figures.   Untransduced