Selection of highly responsive T cell receptors by an analysis combining the expression of multiple markers

Abstract The clinical success of T cell receptor (TCR) gene–transduced T (TCR‐T) cell therapy is expected as one of the next‐generation immunotherapies for cancer, in which the selection of TCRs with high functional avidity (high‐functional TCRs) is important. One widely used approach to select high‐functional TCRs is a comparison of the EC50 values of TCRs, which involves laborious experiments. Therefore, the establishment of a simpler method to select high‐functional TCRs is desired. We herein attempted to establish a simple method to select high‐functional TCRs based on the expression of T cell activation markers using the mouse T cell line BW5147.3 (BW). We examined relationships between the EC50 values of TCRs in interleukin‐2 production and the expression levels of TCR activation markers on BW cells. In TCR‐expressing BW cells stimulated with antigenic peptides, the CD69, CD137, and PD‐1 expression was differentially induced by various doses of peptides. An analysis of TCRs derived from the tumor‐infiltrating lymphocytes of murine melanoma and peripheral blood T cells of hepatocellular carcinoma patients treated with a peptide vaccination revealed that an analysis combining CD69, CD137, and PD‐1 expression levels in BW cells stimulated with a single dose of an antigenic peptide selected high‐functional TCRs with functional avidity assessed by EC50 values. Our method facilitates the section of high‐functional TCRs among tumor‐reacting TCRs, which will promote TCR‐T cell therapy. The stimulation of BW cells expressing objective TCRs with a single dose of antigenic peptides and analysis combining the expression of CD69, CD137, and PD‐1 allows us to select highly responsive TCRs.


| INTRODUC TI ON
T cell receptor gene-transduced T (TCR-T) cell therapy is anticipated as one of the next-generation immunotherapies for cancer patients. 1,2 Tumor-infiltrating lymphocytes (TILs) are a source of tumor antigen-specific CD8 + T cells with TCRs that are used for TCR-T cell therapy. 3 A TCR repertoire analysis by a single-cell analysis [4][5][6] and a comprehensive analysis by deep TCR sequencing using next-generation sequencing combined with a single-cell analysis 7,8 have recently enabled us to easily obtain many tumorspecific TCRs. The development of effective TCR-T cell therapy requires the selection of TCRs with high functional avidity (highfunctional TCRs) among the many tumor-specific TCRs identified to date. 9,10 One widely used approach to select high-functional TCRs is an evaluation of the EC50 values of TCRs, in which T cells are stimulated with various doses of antigenic peptides and their activation is assessed by cytokine production, signal activation, and the expression of activation markers. 9 However, the EC50 analysis requires laborious experiments. Although TCR affinity may also be directly measured by surface plasmon resonance and other methods, [11][12][13] their execution is difficult. Therefore, the development of a simpler method is desired.
We herein attempted to develop a simple method to select highfunctional TCRs. We used BW5147.3 (BW) T cells 14 as reporter cells because they showed the highest sensitivity to the TCR stimulation in our analyses. We assessed the activation of TCR-expressing BW (TCR-BW) cells by analyzing the expression of multiple activation markers (CD137, PD-1, and CD69). The dose dependency of each marker expression differed, which enables us to assess the activation state of TCR-BW cells when stimulated with a single dose of an antigen. Therefore, our method enables the quick and simple selection of high-functional TCRs from many TCRs and will contribute to the promotion of TCR-T cell therapy.

| MATERIAL S AND ME THODS
Mouse TCRs (mTCRs) reacting to the M8 peptide presented by H-2K b were prepared from TILs in a B16F10 tumor that was inoculated into C57BL/6 mice. Human TCRs (hTRs) reacting to the glypican-3 (GPC3) peptide presented by HLA-A*24:02 were prepared from the peripheral blood mononuclear cells (PBMCs) of GPC3 peptidevaccinated patients. mTCRs were transduced into BW cells expressing mouse CD3(mCD3) and mouse CD8αβ, while hTCRs were transduced into BW cells expressing human CD8αβ. To analyze TCR reactivity, COS7 cells expressing H-2K b or HLA-A*24:02 molecules pulsed with antigenic peptides were cultured together with TCR-BW cells for the indicated durations. The secretion of mouse interleukin-2 (mIL-2) by TCR-BW cells was analyzed using an enzyme-linked immunosorbent assay (ELISA), NFAT signal of BW cells by measurements of luciferase activity, and cell surface marker expression on BW cells by FC. To assess the cytotoxicity-inducing ability of TCRs, mouse and hTCRs were transduced into mouse spleen T cells and human PBMCs, respectively. TCR-expressing T cells were cultured together with luciferase-expressing B16F10, luciferase-expressing COS7 cells expressing HLA-A*24:02 pulsed with GPC3 peptides, or HepG2 cells endogenously expressing GPC3 and HLA-A*24:02. Cell viability was analyzed by measuring luciferase activity.
Further details on the materials and methods used in the present study, including reagents, cells, the isolation of TCRs, the expression of TCRs on T cells, the stimulation of TCR-BW cells, analyses of cell activation by ELISA, FC, and cytotoxicity assays are available in Appendix S1.

| Identification of M8-antigen-specific TCRs
To develop an effective method to select high-functional TCRs, we initially attempted to obtain many antigen-specific TCRs. We prepared TILs in B16F10 murine melanoma 15 that were subcutaneously inoculated into C57BL/6 mice. M8 (p15E) is a peptide (KSPWFTTL) derived from an endogenous murine leukemia virus that forms immunodominant epitopes in the context of H-2K b . 16 We stained TILs with a tetramer of the M8 peptide/H-2K b complex together with antibodies to CD8 and CD3, single cell-sorted M8 + CD8 + CD3 + lymphocytes ( Figure S1), and amplified TCR cDNAs from each single cell using multiplex one-step RT-PCR and then analyzed the TCR repertoire. We obtained six T cell populations that contained two or more T cells of the same TCR sequences and 61T cells with unique TCRs ( Figure 1A). To select M8-specific TCRs, we analyzed TCR specificity using two different protocols: (1) We prepared TCR expression vectors and retrovirally transduced TCRs into BW cells that did not express endogenous TCRs. 15,17 TCR-BW cells were cultured with interferon (IFN)γ-stimulated B16F10 cells or M8 peptide-pulsed COS7 cells expressing H-2K b (COS7-H-2K b ), and mIL-2 secretion was analyzed the next day ( Figure S2). (2) We prepared transcriptionally active PCR (TAP) fragments of TCR cDNA and transfected them into human CD8α + Jurkat T cells that did not express endogenous TCR and contained NFAT RE -luciferase gene (Jurkat∆αβ_hCD8α + _luc). TAP fragment-transfected Jurkat∆αβ_hCD8α + _luc T cells were cultured together with M8 peptide-pulsed COS7 cells, and luciferase activity was measured ( Figure S3). 5 We obtained 13 TCRs that specifically responded to the M8 peptide. We expressed them in mouse CD8 + BW cells ( Figure S4) and confirmed their responses to M8 peptidepulsed COS7-H-2K b ( Figure 1B).

| Identification of GPC3 peptide-specific TCRs
We obtained PBMCs from two hepatocellular carcinoma patients who had been involved in clinical trials on the GPC3 peptide vaccine at the National Cancer Center. [18][19][20] PBMCs were stimulated with the GPC3 peptide in the presence of human IL-2. CD137-positive CD8 + T cells were single-cell sorted after their restimulation with the GPC3 peptide ( Figure S5). TCR cDNAs were obtained from single cells by RT-PCR. TAP fragments were prepared and transfected into Jurkat∆αβ_hCD8α + _luc cells. TCR responses to GPC3 peptidepulsed COS7 cells expressing HLA-A*24:02 (COS7-HLA-A*24:02) were analyzed by luciferase activities ( Figure S6). 5 We obtained seven TCRs responsive to GPC3 peptides. We then expressed them in human CD8 + BW cells and examined their responses to GPC3 peptide-pulsed COS7-HLA-A*24:02 cells by analyzing the mIL-2 secretion ( Figure S7). 21 Finally, we obtained five GPC3-specific TCR-BW cells (Figure 2; Figure S8).

| Assessment of the functional avidity of mTCR-BW cells and hTCR-BW cells
We examined the functional avidity of TCR-BW cells. We stimulated mTCR-BW cells and hTCR-BW cells with COS7-H-2K b cells pulsed with various doses of the M8 peptide and COS7-HLA-A*24:02 cells pulsed with various doses of the GPC3 peptide, respectively. TCR-BW cell activation was analyzed by measuring the mIL-2 secretion after a 24-hour culture, and the EC50 (ng/mL) of each TCR was then calculated. EC50 of mTCRs ranged between 0.029 and 1383 ( Figure 3A). EC50 of hTCRs ranged between 3.536 and 179 ( Figure 3B). As assessments of the avidity of many TCRs with a conventional method is laborious, we attempted to develop a simple method to select high-functional TCRs.

| Kinetics of activation markers on the cell surface of BW cells
The expression of several surface markers has been associated with the functional activation of CD8 T cells. [22][23][24] Previous studies reported that the PD-1 expression level was related to the functional avidity of a T cell clone. 25,26 We herein investigated whether surface marker expression levels predicted high-or low-functional TCRs. We mTCR-expressing BW5147.3 (BW) cells were cultured with M8 peptide-pulsed COS7-H-2K b cells, and the mIL-2 secretion was analyzed by ELISA. OTI-TCR-BW cells were used as a negative control. The mIL-2 secretion of mTCR-BW cells stimulated with PMA and ionomycin was shown as a positive control. The significance of differences in the mIL-2 secretion from mTCR-BW cells coincubated with COS7-H-2K b cells pulsed with and without the M8 peptide is indicated. **p < 0.005, ***p < 0.0005, ns: not significant by unpaired, two-tailed t tests. Data are shown as the mean ± SD of triplicate experiments. Representative data of three independent experiments are indicated F I G U R E 2 Responses of GPC3-specific human T cell receptorexpressing BW5147.3 (hTCR-BW) cells. Responses of hTCR-BW cells to GPC3 peptides. hTCR-BW cells were cultured with COS7-HLA-A*24:02 cells pulsed with or without the GPC3 peptide. mIL-2 secretion was analyzed by ELISA. OTI-TCR-BW cells were used as a negative control. hTCR-BW cells were stimulated with PMA and ionomycin and used as a positive control. The significance of the differences in mIL-2 secretion from hTCR-BW cells coincubated with COS7-HLA-A*24:02 cells pulsed with or without the GPC3 peptide is indicated. **p < 0.005, ***p < 0.0005, ns: not significant by  Figure 4C; Figure S9A). When peptide concentrations gradually increased, the upregulated CD69 expression was observed, followed by that of PD-1 and then CD137. The CD69 expression peaked and then decreased at higher peptide concentrations.
We also examined the response of BW cells expressing Epstein-Barr virus (EBV)-specific hTCRs (TCRQ22-BW cells and TCRF39-BW cells) ( Figure S10). TCRQ22 was demonstrated to show higher functional avidity than TCRF39. 6 These cells were cultured with COS7-HLA-A*24:02 cells pulsed with various doses of the BRLF-1 peptide derived from EBV and analyzed as described above. We obtained the results corresponding to Figure 4C (Figure S11).

| T cell receptor signals and marker expression
T cell receptors on BW cells bound the antigenic peptide/MHC complex and delivered signals via mCD3 molecules, which activated BW cells and induced mIL-2 secretion or marker expression. Therefore, we analyzed the relationships between activation signals and mIL-2 secretion or marker expression. We transduced NFAT RE luciferase gene into human and mTCR-BW cells (hTCR-BW-NFAT RE -luc and mTCR-BW-NFAT RE -luc) and assessed TCR-mediated NFAT activation by measuring luciferase activity. When peptide doses increased, dose-dependent elevations in luciferase activity were observed ( Figure S9A). NFAT signals strongly correlated with mIL-2 secretion and CD69 expression ( Figure S9B). Therefore, by comparing the CD137 expression, followed by that of PD-1 and then CD69, we ranked TCR avidity in the order of TCR01-2 4 > TCR02-02 > TCR02-36 ≈ TCR01-10 > TCR02-55, which is consistent with the data shown in Figure 3B.

| Selection of high-functional hTCRs by marker expression on BW cells
At a low concentration of the GPC3 peptide ( Figure 5A Figure S14). The results confirmed that TCRQ22 exhibited higher avidity than TCRF39, which is consistent with the previous data. 6

| Selection of high-functional mTCRs by marker expression on BW cells
We further assessed our method to establish whether highfunctional mTCRs may be selected by combining with an analysis of CD137, PD-1, and CD69 expression. We cultured mTCR-BW cells with COS7-H-2K b cells pulsed with 5 μg/mL of the M8 peptide. After a 24-hour incubation, the MFI of CD137, PD-1, and CD69 expression on mTCR-BW cells were assessed by FC ( Figure 5B; Figure S15 these results show that TCR03-47 exhibited higher avidity than other TCRs, which is consistent with the data shown in Figure 3A, demonstrating that our method is effective for rapidly selecting high-functional TCRs.
Transduction efficiency into splenic T cells was assessed with F I G U R E 5 Selection of high-functional T cell receptors (TCRs) by marker expression on TCR-expressing BW5147.3 (BW) cells. A total of 5 × 10 4 TCR-BW cells were cultured with 5 × 10 4 antigenic peptide-pulsed COS7 cells for a 24-h incubation. The mean fluorescence intensity (MFI) of CD137, PD-1, and CD69 expression was assessed by flow cytometry (FC). A, MFI of markers on human (h)TCR-BW cells. COS7-HLA-A*24:02 cells were pulsed with a high concentration (0.5 μg/mL) of the GPC3 peptide (Left) or a low concentration (0.005 μg/ mL) of the GPC3 peptide (Right). The mean ± SD of three independent experiments is shown. B, MFI of markers on mouse T cell receptor (mTCR)-BW cells. COS7-H-2K b cells were pulsed with a high concentration (5 μg/mL) of the M8 peptide. The mean ± SD of three independent experiments is shown. P values were calculated using Tukey's multiple comparisons test (* p < 0.05, ** p < 0.005, *** p < 0.0005, **** p < 0.0001) the expression of GFP, which varied between 67% and 79% ( Figure S16). We cultured mTCR-transduced T cells together with luciferase-expressing B16F10 (B16F10-luc) cells for 48 hours, and cytotoxic T lymphocyte (CTL) activity was assessed by analyzing the viability of B16F10-luc cells based on luciferase activity ( Figure 6). OTI-TCR-expressing T cells were used as a negative control. A group of high-functional TCRs exhibited higher cytotoxicity than a group of low-functional TCRs. TCR02-B01 with the third highest avidity exhibited the highest cytotoxicity in the highfunctional TCR group.
We obtained a corresponding result as shown in Figure S18.

| DISCUSS ION
Advances in methods for a TCR repertoire analysis and TCR cloning technology have enabled us to obtain many tumor-reactive TCRs from TILs or PBMCs. 5,6 This prompted us to develop a simple method for the identification of high-functional TCRs. In the present study, After a 48-h coculture, the viability of COS7-HLA-A*24:02-luc cells was analyzed by luciferase activity. OTI-TCR-T cells were used as a negative control. Results are obtained from a single experiment representative of at least two independent experiments. Means and SD values from technical triplicate cultures are indicated. Unpaired, two-tailed t tests were performed (*p < 0.05, **p < 0.01, ns: not significant) to antigens ( Figure 4C; Figure S9A). The CD69 expression was rapidly upregulated at a low dose of the antigen and then quickly declined after reaching a peak. The PD-1 expression was upregulated at higher doses of an antigen than CD69 and at lower doses than CD137. The CD137 expression was upregulated at higher doses of the antigen than PD-1 and CD69. These different dose responses in the expression levels of these molecules enabled us to easily assess the avidity of TCRs and select high-functional TCRs with a single dose of an antigen. In TCR-BW cells stimulated with a low dose of the antigen, CD69 was identified as a good marker for the selection of high-functional TCRs. When we used a high dose of the antigen to simulate high-functional TCR-BW cells, the CD137 expression level started to increase, while that of PD-1 peaked. In contrast, the CD69 expression levels rapidly declined after they had peaked.
To evaluate our method, we analyzed M8-specific TCRs derived from TILs in B16F10 melanoma-bearing mice and GPC3-specific TCRs derived from the PBMCs of two cancer patients treated with a GPC3 peptide vaccine. [30][31][32][33] We examined 13 M8-specific mTCRs and five GPC3-specific hTCRs by expressing them on BW cells and assessing marker expression and the mIL-2 secretion. As expected, we selected high-functional TCRs by stimulating TCR-BW cells at a high or low dose of the antigen and analyzing the combination of the CD137, PD-1, and CD69 expression levels, demonstrating that our method easily and simply selected high-functional TCRs.
The present study has several limitations: (1) mTCR02-B01 with the third highest avidity, not mTCR03-47 with the highest avidity ( Figure 5B), exhibited the highest cytotoxicity ( Figure 6). Campillo-Davo et al reported that TCR affinity or avidity does not always correlate with T cell function. 34 Therefore, high-functional TCRs do not always show the highest function when transduced into normal T cells. When TCRs were transduced into splenic T cells, their expression was largely affected by shuffling with endogenous TCRs. [35][36][37][38] Protocols for the knockdown or knockout of endogenous TCRs were recently reported, [39][40][41] and we overcame this limitation by applying these protocols to express the objective TCRs in normal T cells. 42,43 (2) Although the BW cell line showed the highest sensitivity to the TCR-mediated stimulation among the T cell lines examined, sensitivity to the TCR-mediated stimulation of BW cells was lower than that of normal T cells. Therefore, when the avidity of TCRs of tumorresponsive TCRs is too low to analyze using BW cells, TCR avidity cannot be assessed using BW cells. (3) We sometimes experienced that hTCRs on BW cells could not activate the cells even when they were expressed on the cell surface. In this study, we selected seven GPC3-reactive hTCRs using Jurkat human T cells, but TCR01-02 and TCR01-22 could not activate BW cells when they were expressed on BW cells. BW cells express mCD3 molecules. In some case combination of a hTCR and mCD3 molecules may be incompatible and cannot deliver activation signals.
In conclusion, we developed a method to select high-functional TCRs based on an analysis combining the CD137, PD-1, and CD69 expression levels on BW cells. This method allows for the rapid and easy selection of high-functional TCRs from large numbers of tumorspecific TCRs obtained from TILs or peripheral blood T cells and may be applied to TCR-based immunotherapies for tumors. Our method will markedly contribute to advances in TCR-T cell therapies.

ACK N OWLED G M ENTS
We thank Sanae Hirota for technical assistance and Kaoru Hata for secretarial work. We also thank Professor Ellis L. Reinherz and Professor Toshio Kitamura for generously providing us with BW5147.3 cells and Plat-E cells, respectively. The phoenix A cell line was kindly provided by G. Nolan (Stanford University).

FU N D I N G I N FO R M ATI O N
This work was supported by JSPS KAKENHI (grant number JP21K18261, to H.K.).

CO N FLI C T O F I NTER E S T S TATEM ENT
H.K. received research fund from Thyas Co., Ltd. T.N. is a member of the Editorial Board of Cancer Science. Other authors have no conflict of interest.

E TH I C S S TATEM ENTS
Approval of the research protocol by an Institutional Reviewer Bord: A protocol using human samples was approved by the Ethics Review Board of the University of Toyama (R2017016).