Using CombiCells, a platform enabling titration and combinatorial display of cell surface ligands, to study T cell antigen sensitivity by TCRs, CARs, and BiTEs

Understanding how cellular decisions by receptor/ligand interactions at cell/cell interface has been challenging because it is difficult to independently vary the surface density of multiple ligands. Here, we exploit the SpyCatcher/SpyTag split-protein system for rapid combinatorial display of native ligands on cells (Combicells). We use this platform to assess T cell antigen sensitivity and the impact of T cell co-stimulation/co-inhibition receptors. The TCR displayed much greater sensitivity to pMHC than CARs and BiTES did to CD19. While TCR sensitivity was greatly enhanced by CD2 ligand, CAR sensitivity to CD19 was primarily but more modestly enhanced by LFA-1 ligand. Lastly, we show that the PD-1/ligand engagement inhibited T cell activation triggered solely by TCR/pMHC interactions, as well as the amplified activation induced by CD2 and CD28 co-stimulation. The ability to easily produce cells with different concentrations and combinations of ligands should accelerate the study of receptor/ligand interactions at cell/cell interfaces. Graphical abstract One sentence summary Using CombiCells, a platform for the combinatorial display of cell surface ligands, to compare T cell antigen sensitivity mediated by TCRs, CARs, and BiTEs and its dependence on co-stimulation/co-inhibition receptor ligands


Abstract:
Understanding how cellular decisions by receptor/ligand interactions at cell/cell interface has been challenging because it is difficult to independently vary the surface density of multiple ligands. Here, we exploit the SpyCatcher/SpyTag split-protein system for rapid combinatorial display of native ligands on cells (Combicells). We use this platform to assess T cell antigen sensitivity and the impact of T cell costimulation/co-inhibition receptors. The TCR displayed much greater sensitivity to pMHC than CARs and BiTES did to CD19. While TCR sensitivity was greatly enhanced by CD2 ligand, CAR sensitivity to CD19 was primarily but more modestly enhanced by LFA-1 ligand. Lastly, we show that the PD-1/ligand engagement inhibited T cell activation triggered solely by TCR/pMHC interactions, as well as the amplified activation induced by CD2 and CD28 co-stimulation. The ability to easily produce cells with different concentrations and combinations of ligands should accelerate the study of receptor/ligand interactions at cell/cell interfaces.

Introduction
Direct cell-cell communication is a ubiquitous and essential process in multicellular organisms. It is critical during development and tissue maintenance, and underlies the proper functioning of the nervous and immune systems (1). Communication at cellular interfaces relies on diverse families of surface receptors that transduce signals upon recognising their ligands on the surface of other cells. When studying surface receptors that recognise ligands in solution (e.g. G-Protein Coupled Receptors, Receptor Tyrosine Kinases, and Cytokine Receptors), it is trivial to experimentally vary the concentration and combination of soluble ligands. In contrast, it is far more challenging to vary the concentration and combination of cell surface ligands. This technical limitation has hampered our ability to understand cell-cell recognition.
Arguably the most well studied form of cell-cell recognition is T cell antigen recognition. T cells continuously patrol and scan cells throughout the body, seeking abnormal antigens derived from pathogens and mutated proteins produced by cancer cells. T cell activation hinges on whether their T cell antigen receptors (TCRs) bind these antigens, usually in the form of peptides presented on major-histocompatibilitycomplexes (pMHCs). Crucially, the response of the T cell also depends on engagement of other 'accessory' receptors which can enhance or inhibit the response (2,3). Infected or cancerous cells can evade immune recognition by reducing the level of antigen they express on their cell surface. For example, relapses following chimeric antigen receptor (CAR)-T cell therapy are associated with decreases in levels of the target antigen CD19 on the surface of cancer cells (4). In addition, pathogen-infected and cancerous cells can evade immune recognition by changing the levels of ligands to accessory receptors (5)(6)(7). It follows that it is important to be able to investigate how T cell activation is regulated by the concentration of antigens and the combinations of accessory receptor ligands on the target cells.
The accessory receptors CD2, LFA-1, and CD28 are known to enhance T cell responses mediated by the TCR, but their contribution to T cell responses mediated by CARs remains less clear. This is challenging to study as it is difficult to manipulate the surface levels of CAR and accessory receptor ligands. Current methods rely on laborious genetic methods to produce cell lines with desired combinations/surface densities of the required ligands. However, the number of cell lines needed increases exponentially with the number of ligands and surface densities, if all combinations are to be tested, making such experiments impractical. Moreover, the method is susceptible to genetic drift between these cell lines, making it difficult to conclude with certainty that differences observed are actually the result of differences in ligand expression.
Here, we introduce a novel platform enabling the rapid production of cells expressing any combination and concentration of ligands, and we use it to study T cell activation via a native TCR, synthetic CARs, and bi-specific T cell engagers (BiTEs), and the contribution of accessory receptors.

Results
The purified extracellular domain of native ligands fused to Spytag can readily couple to cell surface Spycatcher To enable the combinatorial display of ligands on cells (CombiCells), we reasoned that cell surface expression of the protein Spycatcher, which forms a spontaneous covalent bond with a peptide tag (Spytag) (8), could be used to couple the extracellular domain of purified ligands fused to Spytag (Fig. 1A). Con-sequently, we fused the C-terminus of Spycatcher to the extracellular hinge of human CD52 (hCD52; 7 aa), murine CD80 (mCD80; 20 aa), or a variant of mCD80 that contained fewer residues (mCD80-short; 6 aa). The rationale for coupling the C-terminus of Spycatcher to these short hinges is that it would be expected to maintain a compact conformation bringing Spytag fusion proteins close to the membrane. The CD52 and CD80 hinges are anchored to the cell surface through glycosylphosphatidylinositol (GPI) and a transmembrane domain, respectively. We transduced these surface Spycatchers into CHO-K1 cells and detected expression by coupling a purified fluorescent protein fused to Spytag (Spytag-mClover3, Fig. 1B). A titration of Spytag-mClover3 revealed that the hCD52 hinge surface Spycatcher expressed at the highest level and that saturation was achieved at approximately 1 µM of Spytag-mClover3. Importantly, we confirmed that all surface Spycatchers were mobile at the cell surfacem with diffusion coefficients typical for membrane proteins (Fig. S1). Given its higher expression, we used surface Spycatcher fused to the hinge of hCD52 for subsequent experiments.
T cell activation is known to be controlled in part by the accessory receptors CD2, LFA-1, and CD28, whose ligands are CD58, ICAM-1, and CD86 (or CD80), respectively. To study their individual contributions using surface Spycatcher, a target cell that does not express these ligands is required. Given that CHO-K1 cells are hamster ovary cells, they are not expected to express ligands that bind these receptors, with the exception of ICAM-1, which has been shown to be functional (9). Therefore, we used CRISPR to knockout hamster ICAM-1 before transducing surface hCD52-Spycatcher (Fig. 1C,D). We refer to these CHO-K1 ICAM-1 − hCD52-Spycatcher + as CHO-K1 CombiCells.
We next designed constructs that contained the full extracellular domains of CD58, ICAM-1, CD80, and CD86 fused to a C-terminal Spytag (for coupling to Spycatcher) and Histag (for purification). We produced and purified these ligands and coupled them to CHO-K1 CombiCells before measuring their surface levels using flow cytometry (Fig. 1E). We found that each ligand can be coupled at levels ≳10-fold higher than those found on the T2 cell line, other cell lines, and primary T cells and macrophages (Fig. 1F). Indeed, the absolute number of ligands that can be coupled exceeded ∼ 10 6 per cell (Fig. 1G). We found that coupled ligands had a cell surface lifetime of ≈7 hours detected using ligand-specific or his-tag antibodies (Fig. 1H,  Fig. S2).    (left). An exponential fit is used to determine the mean lifetime (right) and compared using a one-way ANOVA.
The accessory receptor CD2 primarily controls the sensitivity of a pMHC targeting TCR and CAR To study the impact of accessory receptor ligands on T cell antigen sensitivity, we produced purified Spytag-pMHC by refolding HLA-A*02:01 fused to Spytag with β2m and a peptide from the NY-ESO-1 cancer antigen ( Fig. 2A). We performed a preliminary experiment by co-culturing primary human CD8+ T cells expressing the NY-ESO-1 specific 1G4 TCR (10) with CHO-K1 CombiCells loaded with different concentrations of Spytag-pMHC and each ligand. We first confirmed that the surface level of Spytag-pMHC can be varied without impacting the surface level of each Spytag-ligand (and vice versa). We confirmed this to be the case, provided that the total concentration of Spytag-proteins remained below 1 µM, which was the maximum concentration of Spytag-proteins subsequently used (Fig. 2B). We measured T cell activation by surface markers (4-1BB, CD69) and by secreted cytokines (IL-2, IFN-γ, and TNF-α) (Fig. 2C, Fig. S4). We observed the expected increase in T cell activation with increasing concentrations of Spytag-pMHC and with increasing concentration of each Spytag-ligand, consistent with the co-stimulation function of LFA-1, CD2, and CD28. Therefore, this preliminary experiment confirmed that T cells can exploit Spytag-ligands to accessory receptors in recognising Spytag-pMHC in a concentration-dependent manner.
We note that the impact of adding Spytag-ICAM-1 on T cell activation is largely absent on the parental CHO-K1 cell line prior to hamster ICAM-1 knockout (Fig. S3). This underlines the importance of removing endogenous ligands, and demonstrates that T cells can exploit endogenously expressed ligands when recognising pMHC coupled to cell surfaces via Spycatcher/Spytag.
To directly compare the antigen sensitivity of the 1G4 TCR and a CAR, we used the D52N 2nd generation CAR (comprising CD28 hinge, transmembrane, and co-stimulation regions fused to the ζ-chain). This CAR recognises the the same NY-ESO-1 pMHC antigen (11) (Fig. 3A). Both antigen receptors expressed at similar levels (Fig. 3B). We stimulated these cells with CHO-K1 CombiCells presenting different concentrations of antigen, either alone, or in combination with a fixed concentration of one of the co-stimulation ligands. We measured surface markers (4-1BB, CD69), cytokines (IL-2, IFN-g, TNF-a), and TCR/CAR downregulation ( Fig. 3C-E, Fig. S5).
In the case of the TCR, we found that all accessory receptors acted as co-stimulation molecules, but with different quantitative phenotypes. We found that CD2 substantially increased both antigen efficacy (E max ) and sensitivity (P 50 ) for all cytokines and some surface markers, and also increased TCR downregulation. LFA-1 had a more modest effect on antigen sensitivity for surface markers and TCR downregulation, but had no impact on cytokines. Finally, CD28 engagement increased antigen efficacy for IL-2 but had little other impact. These results show that, in expanded human CD8 + T cells, CD2 engagement has a larger impact over a broad range of responses than that of LFA-1 or CD28 engagement. In the case of the CAR, we found a similar qualitative pattern with CD2 imparting the largest co-stimulation effect. However, the quantitative impact was much more modest, with antigen sensitivity improving by 11 and 3.9-fold for 4-1BB and IL-2, respectively, compared to 230 and 46-fold for the TCR. As a result, the fold-difference in antigen sensitivity between the TCR and CAR increased from 30-fold when recognising antigen alone to 300-fold or 120-fold when recognising antigen in the presence of ligands for CD58 or LFA-1, respectively. The lack of any impact of extrinsic CD28 on CAR cytokine production was not unexpected given that it already contained intrinsic CD28 co-stimulation (Fig. 3E).      The accessory receptor LFA-1 primarily controls the sensitivity of CD19 targeting CARs We next investigated the antigen sensitivity of two clinically-approved CAR-T cell therapies targeting the folded antigen CD19 on the surface of B cells, Yescarta and Kymriah. These 2nd generation CARs use the same FMC63 recognition domain fused to either the CD28 hinge, transmembrane, and co-stimulation domains (Yescarta) or CD8 hinge and transmembrane regions and the 4-1BB co-stimulation domains (Kymriah). The current method for studying CAR-T cell antigen sensitivity is to generate panels of cells expressing different levels of the antigen (12,13). However, using a panel of the B cell leukemia Nalm6 cell lines, we found T cell activation was already maximal in response to the clone with the lowest CD19 levels, which was barely detectable by flow cytometry (Fig. S6). We also observed this when studying antigen sensitivity by the TCR finding T cell activation at concentrations of pMHC that were lower (<10 −5 µM, Fig. 2C) than the concentrations required to detect pMHC by flow cytometry (>10 −4 µM, Fig. 2B). This inability to measure antigen surface densities in range needed for measuring T cell sensitivity highlights another advantage of being able to titrate surface antigen levels on cells.
In order to titrate CD19 on target cells, we produced Nalm6 CombiCells by transducing hCD52-Spycatcher into CD19 KO Nalm6 cells (Fig. 4A) and confirmed that purified Spytag-CD19 can readily couple to the cell surface (Fig. 4B). The surface expression of CD19 remained stable for over 24 hours on Nalm6 CombiCells with a lifetime of 49 hrs (Fig. 4C). The surface expression of Spytag-CD19 was less stable on CHO-K1 CombiCells or the U87 glioblastoma cell line expressing hCD52-Spycatcher (Fig. 4C).
When Nalm6 CombiCells were loaded with a range of concentrations of Spytag-CD19 and used to stimulate primary CD8 + T cells expressing either CAR, the antigen sensitivity of Yescarta was 6.3 to 11.5fold higher than Kymriah (Fig. 4D). T cell activation, as measured by 4-1BB surface expression, was detected even when CD19 levels on the Nalm6 surface were too low to detect by flow cytometry (Fig. 4D, black arrow). To investigate the contribution of accessory receptors, we used the CHO-K1 CombiCell assay (Fig. 4E). In contrast to the pMHC-targeting TCR and CAR, we found that the antigen sensitivity of these CD19-targeting CARs was enhanced more by LFA-1 than by CD2 ligands (Fig. 4F). This suggests that CD2 is not being efficiently exploited by the CD19 CARs currently licensed for clinical use.

The antigen sensitivity of the TCR is higher than CARs and BiTEs
We next used the CD19 KO Nalm6 CombiCells to quantify the antigen sensitivity of Blinatumomab, which is an approved BiTE targeting CD19. By including Kymriah and Yescarata, we determined that Blinatumomab performed better than Kymriah and similar to Yescarta in terms of surface 4-1BB, secreted IL-2, and cytotoxicity (Fig. 5A, Fig. S8A). We confirmed that Blinatumomab was not limiting in these experiments as higher concentrations did not impact sensitivity (Fig. S8B).
Taken together, the native TCR appears to display higher antigen sensitivity compared to Blinatumomab and Yescarta, which are more sensitive than Kymriah. The inhibitory PD-1/PD-L1 interaction can inhibit the isolated recognition of pMHC and CD2/CD28 co-stimulation The accessory receptor PD-1 is known to inhibit T cell activation but it is unclear whether it primarily inhibits TCR signalling, CD28 signalling, or both (14)(15)(16)(17). Moreover, it is presently unknown whether PD-1 inhibits co-stimulation by other surface receptors, such as CD2. To investigate this, we used CHO-K1 CombiCells to stimulate CD8 + PD-1 + 1G4 TCR + Jurkat T cells with pMHC alone or with different combinations of ligands to CD28 (CD80), CD2 (CD58) and PD-1 (PD-L1). As expected, ligands for CD28 or CD2 greatly increased T cell activation by pMHC (Fig. 6A-C). In contrast, the PD-1 ligand abolished T cell activation by TCR ligation alone as well as by simultaneous TCR and CD28 ligation. Interestingly, PD-1 ligation also abolished T cell activation by simultaneous TCR and CD2 ligation. These results could not be explained by PD-L1 coupling simply displacing pMHC, CD80, or CD58 because their surface levels were not reduced by coupling of PD-L1 (Fig. S9). These data indicate that PD-1 ligation directly inhibits TCR signalling and therefore, the ability of PD-1 to inhibit CD2 and CD28 co-stimulation may be a result of removing the primary TCR signal and/or its ability to directly inhibit CD2 and CD28 signalling (Fig. 6D).

Discussion
We have developed a new CombiCell platform for studying cell-cell recognition. It adapts the Spycatcher/Spytag split proteins system by expressing a novel membrane-anchored Spycatcher on the surface of cells selected and engineered to lack ligands under investigation. Soluble ligands fused to a membrane-proximal Spytag can readily be coupled to these cells in different combinations and concentrations. This platform, which we call CombiCells, removes a major bottleneck that has been slowing down studies of cell-cell recognition.
CombiCell has several advantages over existing methods, which typically rely on genetic modifications coupled to cell sorting to produce many cell lines with different concentrations and combination of ligands. Firstly, it greatly reduces the number of cell lines. For example, testing just 12 concentrations of antigen with 4 different ligands (e.g. Fig. 4F) would require an impractical 60 cell lines using current genetic methods. With Combicells only one cell line is required. Secondly, cell lines grown independently in culture undergo genetic drift, making it difficult to rule out that observed differences are not the result of such changes. While this could be addressed by creating duplicate cell lines expressing each ligand combination, this would further increase the number of cell lines required. Thirdly, the use of CombiCell allows for generation of cells presenting different ligands within minutes whereas generating cell lines often takes weeks or months. Finally, titration allows ultra-low levels to be displayed on the target cell that are impossible to quantify by flow cytometry. This is crucial when measuring highly sensitive recognition, such as by T cells, which can recognise a single antigen on a target cell (18,19).
To exploit CombiCells, we have focused on T cell activation because the infected or cancerous cells targeted by T cells often modulate expression of surface molecules to evade immune recognition. We show that engagement CD2 had a bigger impact than engagement of LFA-1 or CD28 when T cells recognize pMHC antigens using their TCR. In contrast, LFA-1 had the biggest impact when T cells recognised the cancer antigen CD19 using the clinically approved Yescarta and Kymriah CARs. This is consistent with a recent report showing improved CAR-T cell responses when increasing ICAM-1 expression (7), and suggests that CARs may be under-utilising CD2. Consistent with a previous report, we found that Yescarta achieved higher antigen sensitivity compared to Kymriah (13) and we now report that Blinatumomab (BiTE) performs similarly to Yescarta but that the TCR outperforms both by >10-fold.
Recent studies have suggested that the inhibitory effect of PD-1 involves dephosphorylation of the cytoplasmic tail of CD28 (15,20). We find that PD-1 engagement by PD-L1 can inhibit T cell activation in response to pMHC alone, suggesting that it can also dephosphorylate activatory tyrosines in the TCR signalling pathway (16,17), as originally proposed (21). We also show that PD-1 can inhibit T cell activation enhanced by costimulation through CD2. While CD2 does not contain any tyrosines in its cytoplasmic tail, it has been shown to recruit the tyrosine-containing activatory kinase Lck (22). Our results, taken together with previous reports, are consistent with a model where PD-1 promiscuously inhibits many pathways involving tyrosine phosphorylation (23,24).
While the CombiCell platform has numerous advantages, it also has limitations. Firstly, because these Spycatcher-coupled ligands lack their native membrane/cytoplasmic domains, ligands whose function is influenced by these domains may behave differently. Secondly, even when initial surface densities are matched, the turnover of Spycatcher-coupled ligands is likely to differ from that of native ligands. Since the lifetime of Spytag-protein/Spycatcher complexes ranged from ≈7 hours (Fig. 1H) to >24 hours (Fig. 4C), additional ligand may need to be added for assays of long duration. Finally, this system is not suitable for capturing ligands with multiple transmembrane domains.
By introducing CombiCells we have provided a platform that greatly facilitates the study of receptor/ligand interactions at cell/cell interfaces. We have utilized the platform to compare antigen sensitivity of TCRs, CARs, and BiTEs and the contribution of various accessory receptors to T cell activation, including an inhibitory receptor. This platform can be deployed to examine higher-order combinations of ligands, other surface receptors, and different cell types. CombiCells enable analysis of ligand/receptor interactions at cell/cell interfaces with the convenience hitherto restricted to those studying soluble ligands. This platform should enhance our understanding of how cells integrate signals from diverse surface receptor/ligand interactions at cell-cell interfaces.
Production of Spytag-ICAM-1/CD58/CD86/CD80/PD-L1/CD19: Cells were grown in Expi293™ Expression Medium (ThermoFisher Scientific, A1435101) in a 37°C incubator with 8% CO2 on a shaking platform at 130 rpm. Cells were passaged every 2-3 days with the suspension volume always kept below 33.3% of the total flask capacity. The cell density was kept between 0.5 and 3 million per ml. Before transfection cells were counted to check that cell viability was above 95%, and the density was adjusted to 3.0 million per ml. For 100 ml transfection, 320 µl ExpiFectamine™ 293 Transfection reagent (ThermoFisher Scientific, A14524) was mixed with 6 ml Opti-MEM (ThermoFisher Scientific, 31985062) for 5 min. During this incubation, 100 µg of expression plasmid was mixed with 6 ml Opti-MEM. The DNA was then mixed with the ExpiFectamine™ and incubated for 15 min before being added to the cell culture. One day after transfection 600 µl of enhancer 1 and 6 ml of enhancer 2 was added to the culture flask. The culture was returned to the shaking incubator for 4-5 days for protein expression to take place. Cells were harvested by centrifugation and the supernatant collected and filtered through a 0.22 µm filter. Imidazole was added to a final concentration of 1 mM and PMSF added to a final concentration of 1 mM; 2 ml of Ni-NTA Agarose (Qiagen, 30310) was added per 50 ml of supernatant and the mix was left on a rolling platform at 4°C overnight. The mix was poured through a gravity flow column to collect the Ni-NTA Agarose. The Ni-NTA Agarose was washed three times with 10 ml of wash buffer (50 mM NaH2PO4, 300 mM NaCl, and 5 mM imidazole at pH 8). The protein was eluted with 15 ml of elution buffer (50 mM NaH2PO4, 300 mM NaCl, and 250 mM imidazole at pH 8). The protein was concentrated, and buffer exchanged into size exclusion buffer (25 mM NaH2PO4 and 150 mM NaCl at pH 7.5) using a protein concentrator with a 10,000 molecular weight cut-off. The protein was concentrated down to 500 µl and loaded onto a Superdex 200 10/300 GL (Cytiva, 17-5175-01) size exclusion column. Fractions corresponding to the desired peak were pooled and frozen at -80°C. Samples from all observed peaks were analysed on a reducing SDS-PAGE gel.
For purified Spytag-CD19, SUMO was used to stabilise the protein during production and therefore the HRV 3C Protease Solution Kit was used for SUMO removal (Pierce™, 88946). HRV protease was added to the purified protein at a pre-determined optimum ratio for full cleavage of the HRV site. The mixture was left overnight for full cleave to occur and then 1 ml of Glutathione Agarose (Pierce™, 16100) added for 4 hours to remove the protease. The solution was run through a gravity flow column to collect to SUMO plus protein of interest mixture. This was then added to 1 ml of Ni-NTA Agarose (Qiagen, 30310) and left on a rolling platform at 4°C overnight. The mix was poured through a gravity flow column to collect the Ni-NTA Agarose. The Ni-NTA Agarose was washed once with 10 ml of wash buffer (50 mM NaH2PO4, 300 mM NaCl, and 5 mM imidazole at pH 8). The protein was eluted with 15 ml of elution buffer (50 mM NaH2PO4, 300 mM NaCl, and 250 mM imidazole at pH 8). The protein was concentrated, and buffer exchanged into size exclusion buffer (25 mM NaH2PO4 and 150 mM NaCl at pH 7.5) using a protein concentrator with a 10,000 molecular weight cut-off and frozen in suitable aliquots at -80°C.
Clones were screened using Sanger sequencing after genomic PCR. Specifically, gDNA from outgrown single cell clones was isolated using PureLink Genomic DNA Mini Kit (Invitrogen), amplified in a PCR with fwd primer AGGCATCAGATGGTGGCATTCT and rev primer GGTGTTTGGGGAGGGCAATACT, and submitted for Sanger sequencing. A clone which showed genomic editing was selected for further processing. Next, surface Spycatcher was introduced using high MOI lentiviral transduction, followed by single cell cloning using limiting dilution. The final clone selected showed high expression of surface SpyCatcher and absence of ICAM1 on the cell surface by flow cytometry. The expression of surface Spycatcher was assessed by coupling purified Spytag-mClover and flow cytometry. Specifically, 100k cells were incubated with 10 µM Spytag-mClover in PBS for 1 h at RT in the dark, washed in PBS, and acquired on a flow cytometer. ICAM1 expression was tested using unpurified Y5-3F9 hybridoma supernatant (provided by Vijay Kuchroo and Edward Greenfield). 100,000 cells were incubated with undiluted Y5 supernatant for 30 min on ice in the dark. Cells were washed in PBS and stained with 1:200 anti-mouse Alexa Fluor-488 secondary antibody for 30 min on ice in the dark. Finally, cells were washed and acquired on a flow cytometer. sFCS measurements of diffusion 10 5 CHO-K1 cells expressing surface Spycatcher with different hinges were seeded in 8-well chambered coverslips (µ-Slide 1.5H, ibidi) overnight followed by labeling with 50 nM SpyTag-mClover3 for 30 minutes at 37°C. Cells were washed 2x in PBS and imaged in complete medium. Imaging was performed on a Zeiss LSM 780 inverted confocal microscope (Carl Zeiss) equipped with a 40x C-Apochromat NA 1.2 W FCS objective. mClover3 fluorescence was excited with a 488 nm Argon laser and collected onto hybrid GaAsP detectors (Channel S) using a 488 MBS with the pinhole set to 1 AU. The size of the observation area was calibrated using point-FCS measurements of a dye solution (Alexa Fluor 488, 20 nM) with a known diffusion coefficient (26), yielding an average ω of 214 nm. Diffusion coefficients (D) were then calculated using the equation ω 2 =D×4×t xy where txy is the transit time. Line-scan FCS was performed by switching the ChS to photon-counting mode and data were collected at the basal cell membrane by acquiring a 52pixel line (digital zoom 40x) at maximum scanning speed for 10 5 cycles. Files were saved as .lsm5 files and correlated externally using open-source FoCuS software (27).
Human CD8+ T cells were isolated from leukocyte cones purchased from the National Health Service's (UK) Blood and Transplantation service. Isolation was performed using negative selection. Briefly, blood samples were incubated with Rosette-Sep Human CD8+ enrichment cocktail (Stemcell) at 150µl/ml for 20 minutes. This was followed by a 3.1 fold dilution with PBS before layering on Ficoll Paque Plus (GE) at a 0.8:1.0 ficoll to sample ratio. Ficoll-Sample preparation was spun at 1200g for 20 minutes at room temperature. Buffy coats were collected, washed and isolated cells counted. Cells were resuspended in complete RMPI (RPMI supplemented with 10% v/v FBS, 100 penicillin, 100 streptomycin) with 50U of IL-2 (PeproTech) and CD3/CD28 Human T-activator Dynabeads (Thermo Fisher) at a 1:1 bead to cell ratio. At all times isolated human CD8+ T cells were cultured at 37 and 5% CO2. 1 million T cells in 1ml of media were subsequently transduced on the following day using lentivirus encoding for the 1G4 TCR, Kymriah CAR, or Yescarta CAR, per the section on lentiviral transduction. On days 2 and 4 post-transduction, 1ml of media was exchanged and IL-2 was added to a final concentration of 50U. Dynabeads were magnetically removed on day 5 post-transduction. When using the TCR, T cells were further cultured at a density of 1 million/ml and supplemented with 50U IL-2 every other day. When using CARs, T cells were further cultured at a density of 0.5 million/ml and supplemented with 100U IL-2 every other day. T cells were used between 10 and 16 days after transduction.
Coupling of ligands to CHO-K1 cells 50,000 CHO cells were seeded in a TC-coated 96-well flat-bottom plate and incubated overnight at 37C, 10% CO2. Spytag ligands were diluted to the required concentration in complete DMEM (10% FCS, 1% Penicillin-Streptomycin). Existing media was then removed from CHOs and diluted ligands added in a volume of 50µl, and incubated for 40 or 60 minutes at 37 C, 10% CO2. CHOs were then washed twice with complete DMEM.
Coupling of ligands to Nalm6 cells 30,000 Nalm6 cells were seeded in a TC-coated 96-well round bottom plate and incubated overnight at 37C, 5% CO2. On experiment day, Nalm6 cells were transferred into a TC-coated 96-well V-bottom plate and spun down for 5min at 520g. Spytag ligands were diluted to required concentration in complete RPMI (10%FCS, 1% Penicillin-Streptomycin). Existing media was removed from the Nalm6 cells and the diluted ligands added in a volume of 50µl, and incubated for 40 minutes at 37C, 5% CO2. Nalm6 cells were then washed twice with complete RPMI.
Co-culture assays with TCR or CAR transduced T cells T cells were counted, and washed once in complete RPMI. 50,000 T cells in 200µl complete RPMI were added to CHO cells coupled with ligand in a 96-well flat-bottomed plate or to Nalm6 cells coupled with ligands and transferred into a 96-well round-bottomed plate. The cells were spun at 50g for 1 minute to ensure the T cells settle to the bottom of the plate and make contact with adherent CHO cells. The cells were then incubated at 37C, 5% CO2 for 6 hours (primary T cells) or 20 hours (Jurkat T cells).

Co-culture assays with untransduced T cells and BiTEs
Blinatumomab (BiTE, InvivoGen cat no. bimab-hcd19cd3 ) was resuspended in to a concentration of 100 ug/mL in sterile water and stored in single use aliquots at -20 C until the day of the experiment. Following the coupling of Spytag-CD19, the CD19 KO Nalm6 CombiCells were washed once in media and then seeded at 50,000 cells in 90 uL in 96 well plates. To this, 20 uL of BiTE solution at twice the final concentration (BiTEs were diluted in media) was added and the cells incubated for 30 minutes at 37C. Subsequently, 90 uL of untransduced CD8 + T cells (50,000 cells) were added to give the final indicated BiTE concentration. Effector and target cells were co-cultured for 6 hours. Control cells containing only effector, only target and no BiTE conditions were also seeded in the same volume. 45 minutes before the end of the co-culture 10X cell lysis solution was added to control wells at the appropriate volume to give a final 1X solution, the corresponding volume of sterile water was added to volume correction wells, both for the subsequent cytotoxicity assay. After 6 hours plates were spun briefly at 50 xg for 3 minutes and 100 uL of supernatant carefully removed. 50 uL of the supernatant was used immediately in an LDH release assay using Invitrogen CyQUANT™ LDH Cytotoxicity Assay kits and following the manufacturers protocol. The remaining supernatant was either used immediately or stored at -20 C for subsequent cytokine detection (see below).

Flow cytometry -Detection of ligands
Straight after ligand coupling and subsequent washing, 10mM EDTA was added to the CHO cells to detach them. The cells were transferred to a v-bottom plate and spun for 5 minutes at 500g, 4C. The cells were washed once with PBS-BSA 1% for 5 minutes at 500g, 4C. To detect ligands, fluorescently conjugated antibodies against proteins of interest were diluted in PBS-BSA (1%), at a 1:200 dilution and added at a volume of 50 µl to CHO cells. The cells were resuspended and incubated for 20 minutes at 4C in the dark. The cells were washed twice in PBS, and resuspended in 75 µl PBS, before running on a flow cytometer.

Flow cytometry -Detection of T cell activation
At the end of the stimulation assay, the supernatant was carefully removed and saved for ELISA analysis. 10mM EDTA in PBS was then added to detach the T cells and CHOs. The cells were then aspirated and transferred to a v-bottom plate and washed once in 200µl PBS 1% BSA (500g, 4C, 5 minutes). Antibodies against T cell activation markers were diluted in PBS 1% BSA at a 1:200 dilution. An anti-CD45 antibody was used to selectively stain T cells and distinguish them from CHO cells during flow cytometry analysis. To detect TCR/CAR expression fluorescently-conjugated peptide-MHC tetramers were added to the staining antibodies at a 1:1000 dilution. A viability dye was also added at a dilution if 1:2500 to distinguish live cells from dead cells. 50µl of this staining solution was to the cells, before incubating them for 20 minutes at 4C in the dark. The cells were washed twice in PBS, and resuspended in 75µl PBS, before running on a flow cytometer. Flow cytometry data was analysed using FlowJo (BD Biosciences).

Cytokine detection
IL-2 Human uncoated ELISA kit, TNF-α Human uncoated ELISA kit, IFN-γ Human uncoated ELISA kit, or IL-8 Human uncoated ELISA kit and Nunc MaxiSorp 96-well plates were used according to the manufacturer's instructions. The supernatant from stimulation assays were either undiluted (IL-8) or diluted (all other cytokines) prior to ELISAs. The absorbance at 450 nm and 570nm were measured using a SpectraMax M5 plate reader (Molecular Devices). Data analysis EC 50 is calculated as the concentration of antigen required to elicit 50% of the maximum response determined for each condition individually whereas P X is calculated as the concentration of antigen required to elicit X% of the maximum activation determined by the pMHC alone condition.    µM and data is presented as fold-changes relative to the gMFI of the pMHC alone condition (first column without PD-L1, CD58, or CD80). The protein CD19 fused to Spytag (Spytag-Ctrl) was added so that the total concentration of additional ligand was always 0.2 µM. A one-way ANOVA with Sidak's multiple comparison correction was used to determine p-values. Abbreviations: * = p-value≤0.05, ** = p-value≤0.01, *** = p-value≤0.001, **** = p-value≤0.0001