A T‐cell reporter platform for high‐throughput and reliable investigation of TCR function and biology

Abstract Objective Transgenic re‐expression enables unbiased investigation of T‐cell receptor (TCR)‐intrinsic characteristics detached from its original cellular context. Recent advancements in TCR repertoire sequencing and development of protocols for direct cloning of full TCRαβ constructs now facilitate large‐scale transgenic TCR re‐expression. Together, this offers unprecedented opportunities for the screening of TCRs for basic research as well as clinical use. However, the functional characterisation of re‐expressed TCRs is still a complicated and laborious matter. Here, we propose a Jurkat‐based triple parameter TCR signalling reporter endogenous TCR knockout cellular platform (TPRKO) that offers an unbiased, easy read‐out of TCR functionality and facilitates high‐throughput screening approaches. Methods As a proof‐of‐concept, we transgenically re‐expressed 59 human cytomegalovirus‐specific TCRs and systematically investigated and compared TCR function in TPRKO cells versus primary human T cells. Results We demonstrate that the TPRKO cell line facilitates antigen‐HLA specificity screening via sensitive peptide‐MHC‐multimer staining, which was highly comparable to primary T cells. Also, TCR functional avidity in TPRKO cells was strongly correlating to primary T cells, especially in the absence of CD8αβ co‐receptor. Conclusion Overall, our data show that the TPRKO cell lines can serve as a surrogate of primary human T cells for standardised and high‐throughput investigation of TCR biology.


INTRODUCTION
The genetic replacement of TCRs 1,2 facilitates reprogramming of a T cell's antigen-HLA specificity and offers exciting new prospects for basic research as well as adoptive cell therapy. 3,4 However, especially the identification and indepth characterisation of suitable TCRs for clinical use was so far a tedious process and only a handful of clinical studies with TCR re-directed Tcell products are reported. [5][6][7][8][9] Today, because of continuous improvements in the field of nextgeneration sequencing, high-throughput identification of full ab TCR sequences is no longer a bottleneck. [10][11][12][13] Moreover, advanced bioinformatical analytical tools are developed to gain deep insight into such large TCR repertoire data and to predict antigen-HLA specificity from raw TCR sequences. 14,15 However, a major remaining hurdle is the functional testing of TCR candidates. Earlier studies characterised TCRs by in vitro generation and functional testing of T-cell clones. [16][17][18] Importantly, TCR function is affected by its cellular context, so thatfor instancethe phenotype of a T-cell clone affects TCR functional avidity or even specificity, as previously demonstrated with tumor-infiltrating lymphocytes. 19 Hence, transgenic re-expression of TCRs in a suitable cell line or primary T cells 20 is the most standardised approach to assess TCRintrinsic functionality. However, TCR testing in primary T cells faces an increased degree of variability because of factors such as T-cell activation status, phenotype or donor origin and is also accompanied by high workloads as well as ethical aspects. Hence, the usage of immortalised T-cell clones represents an attractive alternative.
The Jurkat leukemic T-cell line is a widely used model system for the study of TCR function, 21 and we previously developed a triple parameter TCR signalling reporter cell line (TPR) based on the Jurkat line E6.1. 22 These reporter cells have been proven to be highly suitable to evaluate costimulatory pathways and the function of chimeric antigen receptors, [23][24][25] but to date, their potential to evaluate transgenically expressed TCRs in a high-throughput manner that still reflects physiological T-cell biology as seen in primary human T cells had not been tested. To facilitate highly sensitive and unbiased TCR functional characterisation, we introduced two additional modifications in the TPR cell line. First, we introduced the CD8ab co-receptor as it stabilises the TCR-peptide major histocompatibility complex (pMHC) interaction and thereby increases the sensitivity of TCR activation. [26][27][28] Second, since the presence of the endogenous receptor can decrease transgenic TCR functionality [29][30][31] through competition for CD3 molecules 32 and/or formation of mixed TCR dimers, 2,33,34 we performed CRISPR/ Cas9-mediated knockout (KO) of both TCR aand b-chains. Even with these modifications, however, the suitability of such an immortalised cell line for reliable TCR functional testing was not clear. For instance, Jurkat cells are deficient of PTEN 35 which potentially alters TCR functionality in comparison to natural TCR function in primary T cells.
Here, we generated CD8ab +/À endogenous TCRdeficient TPR cell lines (TPR KO -CD8 À and TPR KO -CD8 + ) and comprehensively investigated their suitability for high-throughput TCR functional testing. In total, we transgenically re-expressed 59 human TCRs in TPR KO cell lines and performed an in-depth characterisation of their antigen-HLA specificity and functional avidity. Most importantly, we also performed these experiments in primary human T cells facilitating direct comparison of TCR function between TPR KO cell lines and primary T cells. We observed that a TCR's pMHC-multimer stainability and functional avidity were almost identical in TPR KO cell lines and primary T cells, justifying the usage of our cell line for TCR testing. Furthermore, we document the suitability of TPR KO cell lines for the investigation of TCR biology. Accordingly, we provide further evidence that pMHC-multimer staining is not directly predictive for TCR functional avidity. 36,37 Furthermore, by gathering functional TCR data in the presence or absence of CD8ab, we were able to corroborate previous findings that the CD8ab co-receptor increases peptide sensitivity to a highly TCR-dependent extent 27,28 and that CD8ab dependency inversely correlates with TCR functional avidity. 38,39 Finally, we demonstrate that TPR KO cell lines can be used as the centrepiece of a high-throughput platform for screening of TCRs for clinical use.

Generation of CD8 +/À TCR-replaced Jurkat TCR signal reporter T-cell lines
We previously reported a highly sensitive TCR signal reporter system based on the T-cell line Jurkat E6.1 22 and now aimed to use this cell line for reliable highthroughput evaluation of TCR function. We additionally introduced CD8 aand b-chains (Figure 1a, left panel) to increase the sensitivity of our test system since CD8ab stabilises the TCR-pMHC interaction and promotes TCR-mediated signalling. 27,28,40 As the Jurkat E6.1 cell line expresses an endogenous TCR (as indicated by hTCR and CD3 staining in Figure 1a), we furthermore performed CRISPR/Cas9-mediated KO of TCR aand b-chains ( Figure 1a, right panel). By that, we eliminated potential interactions between endogenous and transgenic TCRs, 2,29,32-34 which would introduce a fundamental source of bias in our test system. KO efficiency was larger than 97% in both cell lines and single CD3-negative cells were sorted on a flow cytometer (Figure 1a, right panel; for the gating strategy, see Supplementary figure 1a).
Subsequently, we validated the full KO of both TCR aand b-chains via polymerase chain reaction (PCR) and Sanger sequencing of the respective CRISPR/Cas9-targeted gene regions (Supplementary figure 1b, c). The resulting TPR KO -CD8 À and TPR KO -CD8 + cell lines were retrovirally transduced with two different A1/pp50 245-253specific TCRs (containing murine constant regions, 41 for all TCR sequences, see Supplementary table 1) in order to validate the function of the TCR signal reporter system. The successful introduction of TCRs was indicated by staining of the transgenic TCR with an antimurine TCR b constant region antibody (mTRBC) and re-expression of CD3 (TCR 14-11 in Figure 1b, see TCR 20-11 in Supplementary figure 2a). Transgenic TCR-expressing TPR KO -CD8 À and TPR KO -CD8 + cell lines were stimulated for 24 h either with peptide-pulsed HLA-A*0101-positive K562 or phorbol 12-myristate 13-acetate (PMA) and ionomycin (Iono). For both cell lines, we observed a peptide-dose dependent NFAT and NFjB reporter activity as well as strong activation via PMA/Iono (TCR 14-11 in Figure 1c, see TCR 20-11 in Supplementary figure 2b). A comparison between TPR KO -CD8 À and TPR KO -CD8 + revealed increased reporter signals in the presence of CD8ab as expected. We further investigated whether the CD8ab co-receptor or the introduced transgenic TCR influences the kinetic of NFAT and NFjB reporter expression, as this would compromise results derived from a snapshot analysis at a certain time point. However, reporter kinetics were highly similar between TCRs as well as between TPR KO -CD8 À and TPR KO -CD8 + cell lines (Figure 1d). We again observed decreased reporter expression in the absence of CD8ab with maximum reporter signal in both cell lines 18 h after stimulation. In summary, we successfully introduced the CD8ab co-receptor and performed CRISPR/Cas9-mediated endogenous TCR-KO in a Jurkat E6.1 based TCR signal reporter cell line.
Moreover, we validated the function of the reporter system with two transgenically expressed TCRs.
pMHC-multimer staining on TPR KO cell lines is reliable and strongly correlates to primary T cells A first crucial step in TCR functional characterisation is the validation of specific target recognition for which staining with pMHC multimeric complexes 42 is a particularly efficient method. However, pMHC-multimer stainings can be a delicate matter. For instance, we observed unsatisfying results with the endogenous TCRdeficient Jurkat 76 cell line (data not shown). Since reliable pMHC-multimer staining would facilitate high-throughput TCR antigen-HLA specificity screening, we compared TPR KO cell lines and primary human T cells in this respect. For this, we introduced TCR 14-11 and TCR 20-11 in primary human T cells and additionally performed KO of the endogenous TCR aand b-chains. We observed highly similar pMHC-multimer staining in our TPR KO cell lines (Figure 2a) in comparison with primary T cells (Figure 2b). Both TCRs, in TPR KO cell lines as well as in primary T cells, showed increased staining intensity in the presence of CD8ab as expected. 27 Interestingly, the two transgenically expressed TCRs in TPR KO -CD8 À or CD4 + primary T cells showed largely different pMHC-multimer staining intensity, indicating differential dependency on the CD8ab co-receptor for pMHC-multimer binding. To further validate the applicability of our cell lines for pMHC-multimer staining and to investigate the TCR-intrinsic ability to bind pMHC-multimer in the presence and absence of CD8ab, we introduced 19 different A1/pp50-specific TCRs in TPR KO cell lines and endogenous TCR-KO primary T cells. For all 19 TCRs, we observed high transduction efficiencies (indicated by mTRBC staining) and highly similar pMHC-multimer stainings between TPR KO cell lines ( Figure 2c) and CD4 + /CD8 + primary T cells (Figure 2d). pMHCmultimer staining of individual TCRs was largely variable, particularly in absence of CD8ab as observed before. 38,39 Quantification of pMHCmultimer staining mean fluorescence intensity (MFI) revealed that CD8ab significantly increases pMHC-multimer staining in TPR KO cell lines ( Figure 2e) and primary T cells (Figure 2f). TCR surface expression was marginally increased in TPR KO -CD8 + but decreased in primary human CD8 + T cells, presumably reflecting slightly different transduction efficiencies and not being generally related to CD8. Weak correlation of mTRBC MFI with pMHC-multimer MFI in TPR KO cell lines (Supplementary figure 3a) and primary T cells (Supplementary figure 3b) indicates that pMHCmultimer stainability is not a mere function of TCR surface expression level but a TCR-intrinsic feature. Furthermore, we observed a large spectrum of different dependencies on the CD8ab co-receptor as quantified by pMHC-multimer MFI fold changes (Figure 2g, h), whereas we did not observe such different dependencies on CD8ab for TCR surface expression (Supplementary figure 3c, d). Most importantly, we observed strong correlations between TPR KO cell lines and primary T cells regarding CD8ab dependency ( Figure 2i) and pMHC-multimer staining intensity (Figure 2j). In case of the latter, the correlation was particularly strong in the absence of CD8ab, indicating that inter-TCR differences in pMHCmultimer staining are to some extent masked by the CD8ab contribution to the TCR-pMHC interaction. In summary, we observed highly reliable pMHC-multimer staining with our TPR KO cell lines that strongly correlates to primary T cells. Using TPR KO -CD8 À and TPR KO -CD8 + cells for pMHC-multimer stainings of 19 individual transgenically expressed TCRs further validated a significant contribution of CD8ab to the stability of the TCR-pMHC complex and revealed a large spectrum of TCR-intrinsic pMHC-multimer stainability, particularly in the absence of CD8ab.
TPR KO cell lines facilitate high-resolution assessment of TCR functionality TPR KO cell lines can thus be used to systematically screen a large number of TCRs for antigen-HLA specificity via pMHC-multimer staining. As a next step, we investigated the suitability of our cell lines for the assessment of TCR functional avidity. For this, we performed antigen-specific stimulation with peptide-pulsed HLA-A*0101 K562 and measured NFAT and NFjB reporter activity after 18 h. We observed a peptide-dose dependent reporter response in both TPR KO cell lines for two individual A1/pp50-specific transgenically expressed TCRs ( Figure 3a). As observed before (Figure 1c), the reporter signal was increased in the presence of CD8ab, but yet again, to a different extent between TCRs. To investigate TCR functionality in our TPR KO cell lines in more detail, we performed peptide titrations with 19 A1/pp50-specific TCRs. TPR KO figure 4g) were not strong, mainly because of the small functional differences between TCRs in the presence of CD8ab but also indicating TCR-intrinsic CD8ab co-receptor dependency. Interestingly, the correlation of CD8ab dependency to functional avidity revealed an inverse correlation that was particularly strong in TPR KO -CD8 À (Figure 3h). In summary, TPR KO cell lines facilitate TCR functional characterisation with high resolution and low technical and/or biological variability. TCR-intrinsic differences in functional avidity are particularly visible in the absence of CD8ab. Furthermore, co-receptor dependency inversely correlates to functionality. Hence, low avidity TCRs disproportionally benefit from CD8ab, whereas high avidity TCRs show only little additional gain in peptide sensitivity.

TCR functional avidity determined in TPR KO cell lines strongly correlates to primary T cells
We have demonstrated that TPR KO cell lines can be used for large scale assessment of TCR     functional avidity. However, we were concerned whether these TCR functionality data accurately reflect data obtained with primary human T cells. While the Jurkat cell line represents a generally accepted model system for investigation of T-cell activation and TCR signalling, there might also be TCR function affecting differences between this immortalised cell line and primary T cells, such as a reported PTEN deficiency 35 . Therefore, our goal was to systematically compare TCR functionality in TPR KO cell lines to primary human T cells. For this, we introduced the same 19 A1/pp50-specific TCRs (shown in Figure 3) into endogenous TCR-KO primary CD4 + /CD8 + T cells and performed intracellular cytokine staining of interferon gamma (IFNc) and tumor necrosis factor alpha (TNFa) upon antigen-specific stimulation with peptide-pulsed HLA-A*0101 K562. In general, investigation of transgenically expressed TCRs in primary T cells revealed highly similar relations between TCR functional avidity and CD8ab coreceptor contribution as observed in TPR KO cell lines: peptide sensitivity and E max were increased in CD8 + compared to CD4 + primary T cells (  figure 5g) were not strong as observed in TPR KO cells; and IFNc and TNFa EC 50 inversely correlated to CD8ab co-receptor dependency, which was particularly strong in CD4 + T cells (Figure 4h).
Direct comparison of NFAT and NFjB EC 50 values measured in TPR KO cell lines with IFNc and TNFa EC 50 values measured in CD4 + /CD8 + primary T cells revealed a surprisingly strong correlation, particularly in the absence of CD8ab co-receptor (Figure 5a), indicating that inter-TCR differences are masked by CD8ab contribution. CD8ab coreceptor dependency of functional avidity was also strongly correlating between TPR KO cell lines and primary T cells (Figure 5b). We further related functional avidity data to pMHC-multimer staining data and did not observe a correlation in both TPR KO figure 6b), neither for CD8 + nor for CD8 À cells. Accordingly, CD8ab co-receptor dependency of functional avidity and CD8ab co-receptor dependency of pMHC-multimer staining also did not correlate in TPR KO cell lines (Supplementary figure 6c) and primary T cells (Supplementary  figure 6d). These findings generated with a plethora of different TCRs systematically side-byside are in line with previous reports that document no or at most a very limited correlation between pMHC-multimer stainability and TCR functionality. 36,37 Interestingly, these data further indicate that CD8ab contributes to pMHC-multimer staining and functional avidity via different mechanisms.

cell lines (Supplementary figure 6a) and primary T cells (Supplementary
Most importantly, we show that TCR functional avidity in TPR KO cell lines strongly parallels TCR functional avidity in primary T cells. Hence, TPR KO cell lines can be used as a surrogate of primary T cells, which facilitates a high-throughput, standardised and reliable characterisation of TCR functional avidity. Furthermore, our data on the relations of CD8ab co-receptor to pMHC-multimer staining and functional avidity illustrate the suitability of our TPR KO cell lines for investigation of TCR biology in general.

TPR KO cell lines as the centrepiece of a highthroughput TCR screening platform
In order to validate the suitability of our TPR KO cell lines for high-throughput and reliable determination of TCR antigen-HLA specificity and functionality, we tested our system with 38 TCRs that were initially isolated by flow cytometry sorting of A2/pp65 495-593 pMHC-multimer + CD8 + T cells. First, we performed retroviral transduction of all 38 TCRs into TPR KO -CD8 + cells to determine TCR surface expression and antigen-HLA specificity via pMHC-multimer staining. 30 TCRs could be restained with pMHC-multimer, whereas seven TCRs did not stain with pMHC-multimer (TCRs 13-4, 56-10, 59-10, 67-8, 70-8, 71-8, and 79-14) and one TCR was not expressed at all on the cell surface (TCR 58-10) (Figure 6a). For TCR 13-4, we confirmed the lack of antigen-HLA specificity in primary T cells (Supplementary figure 7). Having identified 30 A2/pp65-specific TCRs, we subsequently determined their functional avidity. In order to streamline the measurement of 30 TCRs upon stimulation with six different peptide concentrations in triplicates (equals 540 samples),  we performed multiplexing via a CD45 antibody barcoding approach. Using combinations of three differently fluorochrome-labelled CD45 antibodies, eight individual samples receive a unique colour barcode and can thereby be pooled within one sample (Figure 6b). The sample number was thereby reduced to 72. Usage of additional CD45 antibodies with different fluorochrome labels could have easily further decreased this number. Quantification of NFAT ( Figure 6c) and NFjB (Supplementary figure 8a) EC 50 values revealed a large spectrum of different TCR functional avidities, particularly in absence of CD8ab as observed before with A1/pp50-specific TCRs (Figure 3b and Supplementary figure 4a). Based on NFAT EC 50 values measured in TPR KO -CD8 À cells, we selected eleven TCRs, covering the whole avidity spectrum (Figure 6c, marked in red colour), for TCR re-expression and functional characterisation in primary human T cells. Again, we observed a large spectrum of IFNc ( Figure 6d) and TNFa (Supplementary figure 8b) EC 50 values in CD4 + T cells, whereas this diversity was decreased in CD8 + T cells. Between TPR KO cell lines and primary T cells, the functionality of these eleven A2/pp65-specific TCRs correlated well in absence, but not in presence of CD8ab (Figure 6e). Finally, we also compared pMHC-multimer staining of these TCRs in TPR KO  CD8ab as observed with A1/pp50-specific TCRs (Figure 2j). In summary, we here provide proof-ofconcept for the suitability of our TPR KO cell lines for high-throughput and reliable screening of TCR antigen-HLA specificity and functional avidity.

DISCUSSION
The functional characterisation of TCRs is most widely performed after transgenic re-expression in primary T cells. Variability in primary T cells because of phenotype, activation status, or donor origin can affect TCR function and bias results. Hence, a cell line that provides TCR function close to primary T cells would enable more standardised testing as well as simplify the whole process because of cell lines' easy handling and almost unlimited proliferative capacity. The urgent need for such a cell line is highlighted by various publications that proposed different cellular platforms for TCR testing. [43][44][45][46][47] Here, we propose an advanced Jurkat E6.1based TCR signal reporter system that is unbiased by endogenous TCR expression. Our study, which analysed 59 different human TCRs, isto our knowledgethe first to comprehensively compare TCR function in a cell line with primary human T cells. As TCR function was closely parallel to primary T cells, our TPR KO cell lines proved highly suitable for functional characterisation of individual TCRs and also for the investigation of TCR biology in general. By relating functional avidity to pMHC-multimer staining data, both in TPR KO cell lines and primary T cells (Supplementary figure 6), we validated that pMHC-multimer staining intensity is not predictive for functionality, 36,48 highlighting the importance of functional testing for the identification of suitable TCRs for clinical use. Further, we confirmed previous findings that the CD8ab coreceptor increases a TCR's peptide sensitivity to a highly differential TCR-dependent extent 27,28 and CD8ab co-receptor dependency inversely correlates with functional avidity. 38,39,49 The latter implicates that measured TCR functional avidity in absence of CD8ab might more directly reflect the structural avidity of a TCR to its cognate pMHC. We further observed a disparity between CD8ab dependency of pMHC-multimer staining and of TCR functional avidity, indicating the presence of two different mechanisms of CD8ab contribution to pMHC-multimer binding and antigen-specific TCR activation. Our TPR KO cell lines could be used as a tool to investigate this more closely.
Whereas TCR sequencing [10][11][12][13] and antigen-HLA specificity prediction algorithms 14,15 are in constant progress, validation of TCR specificity and function remains a bottleneck. On this aspect, recently reported protocols for high-throughput direct cloning of TCRs for transgenic re-expression represent major progress for large scale TCR reexpression. 12,13,50 We here document highly reliable pMHC-multimer staining on our TPR KO cell lines demonstrating their suitability for large scale antigen-HLA specificity screening approaches. For instance, this enables re-expression of large combinatorial libraries of TCR aand b-chains in our TPR KO cell lines for high-throughput screening of antigen-HLA specificities of interest. Furthermore, we have demonstrated that TPR KO cell lines facilitate a high-throughput functionality screening of TCRs with high sensitivity and reliability. Hence, TPR KO cell lines enable the generation of large datasets connecting TCR sequence, antigen-HLA specificity, and function to an unprecedented extent. This would be a substantial contribution to the development of improved algorithms for antigen-HLA specificity and probably also functionality prediction from raw TCR sequence data. 14,15,51 Fast sequencing of TCR repertoires in combination with such reliable prediction algorithms has the potential to revolutionise patient-individualised adoptive T-cell therapy. Determination of TCR functionality in the presence and absence of CD8ab enables identification of largely CD8ab co-receptorindependent TCRs, which could be of particular interest for clinical application. On the one hand, CD8ab-independent TCRs would maintain their functionality in T-cell products despite largely variable CD8ab expression. 39 On the other hand, it was shown that CD4 + T cells expressing an MHC class I-restricted TCR provide important additional TCR functions, such as increased IL-2 help, and thereby contribute to an increased anti-tumor response. [52][53][54] Hence, CD8ab co-receptor-independent TCRs would represent ideal candidates for such an approach.
In summary, we here propose a Jurkat-based TCR signal reporter cell line for testing of TCR specificity and functionality unbiased by endogenous TCR expression. TCR functional avidity of 30 individual TCRs in our TPR KO cell lines was strongly correlating to primary human T cells, highlighting the suitability of our cell line for highly reliable investigation of TCR function and biology. Hence, this platform represents a valuable tool for the characterisation and selection of TCR candidates for clinical use and also facilitates the generation of large TCR functionality datasets for the development of prediction algorithms.
Written informed consent was obtained from peripheral blood mononuclear cell (PBMC) donors, and usage of the blood samples was approved according to national law by the local Institutional Review Board (Ethikkommission der Medizinischen Fakult€ at der Technischen Universit€ at M€ unchen). The study conforms to the standards of the Declaration of Helsinki.

TCR identification
PBMCs of CMV-seropositive, healthy donors were stained with respective pMHC-multimer that was individually conjugated with two different fluorophores to achieve reliable double pMHC-multimer staining. Single cells positive for CD8, CD62L, CD45RO, and both pMHC-multimer conjugates were sorted in a 384-well plate and stimulated with 10 µg mL À1 plate-bound anti-CD3 and anti-CD28 each. RPMI medium was supplemented with 200 IU mL À1 IL-2 and 5 ng mL À1 IL-15. Single cell-derived clones were harvested between days 7 and 14 after sorting. TCRs were amplified via TCR-SCAN RACE PCR 55 and subsequently sequenced on the Illumina MiSeq platform. TCR nomenclature represents a consecutive numbering with no meaning for the here presented data.
TCR DNA template design DNA templates were designed in silico and synthesised by GeneArt (Life Technologies, Thermo Fisher Scientific) or Twist Bioscience (San Francisco, California). DNA constructs for retroviral transduction had the following structure: Human Kozac sequence 56 followed by TCR b (including a murine TCR b constant region (TRBC) with additional cysteine bridge 41,57,58 ), followed by P2A and followed by TCR a (including a murine TCR a constant region (TRAC) with additional cysteine bridge 41,57,58 ), cloned into pMP71 vectors (kindly provided by Wolfgang Uckert, Berlin).
packaging cell line. For the production of retroviral particles, RD114 cells were transfected with pMP71 expression vector (containing the TCR construct) by calcium phosphate precipitation. Virus supernatant was harvested after 72 h and subsequently coated on retronectin-treated (TaKaRa; Kusatsu, Japan) well plates. Bulk PBMCs were activated for two days with CD3/CD28 Expamer (Juno therapeutics a Bristol-Myers Squibb Company), 300 IU mL À1 IL-2, 5 ng mL À1 IL-7 and 5 ng mL À1 IL-15 per mL of RPMI for 1 9 10 6 T cells. Expamer stimulus was removed by incubation with 1 mM D-biotin. 1 9 10 5 mL À1 TPR cells were seeded in a 24-well plate two days before transduction. Activated T cells or TPR cells were transduced via spinoculation on virus-coated plates. TCR transduction occurred 15 min after CRISPR/Cas9-mediated TCR-KO editing of T cells.

Antigen-specific activation and TCR signalling
TCRs were introduced into the TPR KO cell lines via retroviral transduction. Antigen-specific stimulation was performed using irradiated (80 Gy) and peptide-pulsed (10 À9 M, 10 À8 M, 10 À7 M, 10 À6 M, 10 À5 M, 10 À4 M) K562 cells (retrovirally transduced to express the MHC class I molecule of interest). Effector and target cells were co-cultured in a 1:5 ratio for 18 h. Subsequently, NFAT-GFP and NFjB-CFP reporter expression was analysed on a flow cytometer.

Sanger sequencing for KO validation
Genomic DNA was extracted (Wizard SV Genomic DNA Purification System, Promega; Madison, Wisconsin) from flow-sorted CD3-negative TPR cells. PCRs were performed to amplify the intended CRISPR/Cas9-mediated cutting sites within the first exon of TRAC as well as the first exon of TRBC1/2. Purified PCR products were Sanger sequenced (Eurofins Genomics, Ebersberg, Germany).

Flow cytometry
Acquisition of FACS samples was done on a Cytoflex (S) flow cytometer (Beckman Coulter). Flow sorting was conducted on a FACSAria III (BD Bioscience) or MoFlo Astrios EG (Beckman Coulter).