Antigen‐specific TCR‐T cells from Rag2 gene‐deleted pluripotent stem cells impede solid tumour growth in a mouse model

Abstract The technology of adoptive transfer of T‐cell receptor (TCR) engineered T cells is wildly investigated as it has the potential to treat solid cancers. However, the therapeutic application of TCR‐T cells is hampered by the poor quality derived mainly from patients' peripheral blood, as well as heterogeneous TCRs caused by the mismatch between transgenic and endogenous TCRs. To improve the homogeneity, antigen‐specificity and reduce possible autoreactivity, here we developed a technique to generate antigen‐specific T cells from Rag2 gene‐deleted pluripotent stem cells (PSCs) and further measured their anti‐tumour efficacy. PSCs were first targeted with OT1 TCR into the Rag2 locus to prevent TCR rearrangement during T‐cell development. The engineered PSCs were then differentiated through a two‐step strategy, in vitro generation of haematopoietic progenitor cells, and in vivo development and maturation of TCR‐T cells. Finally, the response to tumour cells was assessed in vitro and in vivo. The regenerated OT1‐iT displayed monoclonal antigen‐specific TCR expression, and phonotypic normalities in the spleen and lymph node tissues. Importantly, the OT1‐iT cells eliminated tumour cells while releasing specific cytokines in vitro. Furthermore, adoptive transfer of OT1‐iT cells suppresses solid tumour growth in tumour‐bearing animals. Our study presents a novel and straightforward strategy for producing antigen‐specific TCR‐T cells in vivo from PSCs, allowing for allogeneic transplantation and therapy of solid tumours.

producing antigen-specific TCR-T cells in vivo from PSCs, allowing for allogeneic transplantation and therapy of solid tumours.

| INTRODUCTION
T cells engineered to express TCRs have shown encouraging results in the therapy of melanoma and sarcoma. [1][2][3][4][5] TCR-T cells, unlike CAR-T cells, can recognize particular antigens on or within of cancer cell membranes as peptides that attach to major histocompatibility complex (MHC) molecules, allowing them to recognize a far broader spectrum of antigens. 6 However, a variety of obstacles have hindered the therapeutic application of TCR-T cells. For instance, transgenic TCRs compete with endogenous TCRs, which leads to dilution of tumour specific TCRs and decreased T cell avidity and anti-tumour efficacy. TRAC or TRBC knocked off using ZFN have strong anti-tumour activity. 11 With the development of gene editing technology in recent years, it is now simple to delete TRAC or TRBC using TALEN or CRISPR-Cas9. 12 Pluripotent stem cells (PSCs) provide an unlimited source of cells that can be genetically manipulated and differentiated into more specialized T cells. [16][17][18] Minagawa et al. recently reported the generation of TCR-stabilized CD8αβ T cells from T-iPSCs via knockout of recombination activating genes 2 (Rag2). 19 The Rag1 or Rag2 gene, which mediates the TCR V(D)J recombination during CD4 and CD8 double negative and CD4 and CD8 double positive stages, might be the ideal targets for reducing endogenous TCRs. 20 However, the initial T-iPSCs were obtained by laborious screening of antigen-specific T-cell clones and challenging reprogramming them into T-iPSCs. The production and selection of T-iPSCs are timeand money-consuming processes.
It would be more practicable to develop an approach for directly producing transgenic TCR-T (TgTCR-T) cells from engineering PSCs.
Previously, our groups reported a novo method to generate T cells in vivo by overexpressing particular transcription factors Runx1 and Hoxa9 in PSCs. 21,22 Here, we used CRISPR-Cas9 technology to further target transgenic TCR into the Rag2 locus of PSCs. We validated that PSCs' capacity to differentiate into T cells was not impeded by genetic engineering. Importantly, we showed that this approach can effectively generate a considerable number of desired tgTCR-T cells in vivo. These PSC-derived tgTCR-T cells exhibited high TCRmediated specificity as well as a powerful anti-tumour activity. Our founding might provide an efficient strategy to obtain limitless and stable TCR-T cells from PSCs that have been genetically targeted to a tumour antigen of interest.
Tumour size (lengthÂwidth) was measured every 2 days. Mice suffered from heavy tumour burdens (tumour size larger than 400 mm 2 ) were euthanized for ethical consideration. The survival time of tumour mice was recorded.

| Statistics
All data presented in the figure panels were performed using GraphPad Prism and based on at least three replicated samples.
Two-tailed Student's t-test, one-way ANOVA, and log-rank tests were used by SPSS software to calculate statistical significance.

| Targeted integration of a TCR into Rag2 locus of PSCs
Our previous studies found that ectopic expression of Runx1 and Hoxa9 in mouse PSCs (iR9-PSCs) efficiently induced the differentiation of haematopoietic progenitor cells (iHPCs) in vitro, and then successfully generated mature T lymphocytes in vivo. 21,22 To further produce tumour antigen-specific CD8 T cells, we selected OVAantigen TCR (OT1, MHC-I restricted, recognizing OVA 257-269 ), a well-characterized CD8 TCR. To prevent the production and rearrangement of endogenous TCRs, we used the CRISPR-cas9 technology to knockout recombination activating genes 2 (Rag2) and knockin OT1 under the control of CAG promotor of iR9-PSCs (referred to as Rag2 À/À -OT1-iR9-PSCs) ( Figure 1A). According to Rag2 gene sequencing, one allele of the gene had 186 base pair deletions, while the other allele had an OT1 exchange ( Figure 1B). The 186 bp deletion on one allele created a mutant protein with 62 amino acids shortened, which is crucial for the structure and catalytic activity of the Rag complex, since the minimal core of Rag2 has been considered to be residues 1-351 of 527. 20,24,25 PCR amplification and gel electrophoresis were used to validate the OT1 exchange on another allele ( Figure 1C). The intracellular costaining of TCR Vα2 and Vβ5 in Rag2 À/À -OT1-iR9-PSCs was then carried to identify OT1 expression. As shown in Figure 1D, Rag2 À/À -OT1-iR9-PSCs demonstrated a high level of OT1 expression.

| Rag2 À/À -OT1-iT cells inhibited the growth of specific solid tumour in vivo
To explore the therapeutic potential of these PSC-derived OT1-iT cells in vivo, we used a B16F10-OVA tumour model in SCID mice. In this model, 1.5 Â 10 5 B16F10-OVA tumour cells were subcutaneously injected into SCID mice to produce tumour growth. On Day 12, Rag2 À/À -OT1-iT cells (1.0 Â 10 7 /mouse) which were derived from the spleens of Rag1 À/À recipients transplanted with day21-iHPCs for 6 weeks, were retro-orbitally to these tumour-bearing mice ( Figure 4A). Every 2 days, tumour size was monitored by measuring the length and width. On Day 18 after B16F10-OVA injection, the tumour sizes of Rag2 À/À -OT1-iT-treated mice were less than 100 mm 2 , but those of PBS-treated tumour-bearing mice were over 200 mm 2 (n = 5, p < 0.001) ( Figure 4B), indicating that Rag2 À/À -OT1-iT cell therapy inhibits tumour growth significantly. We also observed that Rag2 À/À -OT1-iT cell treatment dramatically improved tumour-bearing mice survival as compared to PBS-treated animals ( Figure 4C). The phenotype of the infiltrating T cells in the tumour and spleen was subsequently determined. As shown in Figure 4D, primary T cells displayed the naive CD8 T cell phenotype: CD44 low CD62L high CD69 À CD25 À . After tumour stimulation in vivo, these tumour-infiltrating Rag2 À/À -OT1-iT cells exhibited the typical effector cytotoxic T cells phenotype: CD44 high CD62L low CD69 + CD25 + , whereas the splenic Rag2 À/À -OT1-iT cells showed the memory phenotype (CD44 high CD62L high CD69 À CD25 À ). Simultaneously, to evaluate the efficacy of PSC-derived Rag2 À/À -OT1-iT cells and the physiological OT1-T cells from OT1 transgenic mice, we performed a detailed comparison between those two cells. Surprisingly, Rag2 À/À -OT1-iT cells had a similar anti-tumour response and phenotypic status to OT1-T cells ( Figure 4B-D). Overall, the findings indicate that iT cells regenerated from TCR-engineered PSCs have excellent anti-tumour activity against solid tumours due to better tumour infiltration and memory capacity.

| DISCUSSION AND CONCLUSION
TCR-T clinical studies have yield impressive outcomes. However, the supply of therapeutic T cells is a major bottleneck. TCR-T cells generated from PSCs might be employed to address the issue. Many research groups have struggled to generate antigen-specific CD8αβ T cells from T-iPSCs, 17,18,[27][28][29] but here we developed a straightforward strategy for producing TCR-iT cells from PSCs by CRISPR-based TCR knockin to the Rag2 locus. Due to the absence of TCR rearrangement during PSCs differentiation, the regenerated TCR-iT cells were pure and superior at antigen recognition. Similar to TCR-T cells from transgenic mice, these TCR-expressing PSC-derived iT cells inhibit tumour growth in a xenograft model (Figure 4). In addition, our findings reveal that TCR-iT cells produced from PSCs present not only an effector phenotype in the tumour tissue, but also a memory phenotype in the spleen (Figure 4), which is important for immunological surveillance. More interestingly, direct transplantation of iHPCs derived from Rag2 À/À -OT1-iR9-PSCs may also effectively suppress the development of tumours and significantly prolong the survival time of tumour-bearing mice, indicating that iHPCs can swiftly give rise to iT cells in vivo ( Figure 5). Our findings showed that either TCR-iT cells or iHPCs can efficiently inhibit tumour growth in tumour-bearing mouse model. TCR-iT cells can eliminate tumours more quickly than iHPCs since they are more mature than iHPCs. In addition, TCR-iT cells may be safer than iHPCs, because of their homogeneity and shorter lifespan, which may reduce the risk of generating unknown pluripotency cells with tumour potential. However, iHPCs may be better suited for early tumour suppression, and their clinical application requires more research.
Although the potential of PSC-derived TCR-T in cancer therapy has been acknowledged, how to avoid its allogeneic rejection is still a technical challenge. Graft versus host disease (GvHD) is mainly caused by graft T cell recognition of host peptide-HLA complexed through TCR. 30 To eliminate the TCR-mediated GvHD, the main method is to disrupt the αβ chain of TCRs by targeting TRAC or TRBC. Because TCRs are essential for T-cell development in thymus, such approaches can only be used on mature T cells. Here, we described a simple method for obtaining desired TCR-iT cells from PSCs using Rag2 sitespecific recombinant. We validated that allogeneic iT cells engineered with TCR sequence in Rag2 locus have normal T-cell development, maturation or migration since the TCR-iT cells show normal tissue distribution and adaptive immunity as OT1-T cell from transgenic mice.
Furthermore, the TCR-iT cells resulted in no GvHD after transplanting Rag2 À/À -OT1-iR9-iT cells (H-2 b ) to allogenic SCID mice (H-2 d ). Our research established a method to produce TCR-iT cells from mouse PSCs, paving the way for human TCR-T-cell production. However, there are still certain issues that need to be solved for human TCR-Tcell regeneration. For example, more study is required to confirm whether the synergistic expression of human RUNX1 and HOXA9 can efficiently induce the differentiation of human PSCs to regenerate human T cells. Furthermore, it is necessary to perform this approach in the humanized animal model transplanted with human foetal thymus during the development of T cells.
In our study, we combined PSC with TCR engineering technologies to offer a potential new supply of 'off-the-shelf' TCR-T cells. By targeting disruption of B2M or HLA genes, PSC-derived TCR-T can be further modified to reduce host versus graft disease. 31,32 If successful, these strategies may allow allogeneic transplantation of PSCproducing TCR-T cells.

AUTHOR CONTRIBUTIONS
Jinyong Wang and Hongling Wu conceived and supervised the study.