Generation of effective and specific human TCRs against tumor/testis antigen NY-ESO-1 in mice with humanized T cell recognition system

Abstract

Generation of high avidity T cell receptors (TCRs) reactive to tumor-associated antigens (TAA) is impaired by tolerance mechanisms, which is an obstacle to effective T cell therapies for cancer treatment. NY-ESO-1, a human cancer-testis antigen, represents an attractive target for such therapies due to its broad expression in different cancer types and the restricted expression in normal tissues. Utilizing transgenic mice with a diverse human TCR repertoire, we isolated effective TCRs against NY-ESO-1_157-165 restricted to HLA-A*02:01. We compared the functions of the murine-derived TCR with human-derived TCRs and an affinity matured TCR, using in vitro co-culture and in vivo adoptive T cell transfer in tumor-bearing mice. Alanine scan, x-scan, LCL assay were employed to address the cross-reactivity of the NY-ESO-1_157-165 specific TCRs. We also used human tissue cDNA library and human primary cells to assess the safety of adoptive T cell therapies targeting NY-ESO-1 antigen in the clinic. One of the murine-derived human TCRs, TCR-ESO, exhibited higher functional avidity compared to human-derived NY-ESO-1_157-165 specific TCRs. TCR-ESO appeared to have similar efficiency in antigen recognition as an in vitro affinity-matured TCR, TCR 1G4-α95LY, which was applied in clinical trials. TCR-ESO showed little cross-reactivity, in contrast to TCR 1G4-α95LY. Our data indicate that highly effective TCRs against NY-ESO-1 are likely deleted in humans due to tolerance mechanisms, and that the TCR gene loci transgenic mice represent a reliable source to isolate effective and highly-specific TCRs for adoptive T cell therapies.

Keywords: HLA-A*0201; NY-ESO-1; TCR engineering; adoptive T cell therapy; humanized mice.

Figure 1

(A) IFNγ intracellular staining of blood cells of NY-ESO-1 immunized ABab-A2 mice. ABab-A2 mice were gene gun immunized and boosted with NY-ESO-1 full length cDNA. Blood was drawn 7 days after the last boost and cells re-stimulated for 14 hours with MAGE-A1₂₇₈ (irrelevant peptide), NY-ESO-1_{157-165(9CtoA)} peptide or anti-mouse CD3/CD28 beads and stained intracellularly for IFNγ production. One representative experiment of 21 immunized mice is shown. (B) Re-expression of TCR1, TCR-ESO and TCR3 in human PBMCs. Human PBMCs were transduced with NY-ESO-1_157-165 TCRs encoding retrovirus and stained with NY-ESO-1_157-165/HLA-A2 tetramer. (C) Peptide titration of TCR1, TCR-ESO and TCR3 transduced huPBMCs. Different concentration of NY-ESO-1_157-165 peptide were loaded onto T2 cells as targets for TCR transduced huPBMCs and cultured overnight. IFNγ levels were then measured by human IFNγ ELISA. One-site specific binding curves were calculated on the normalized IFNγ secretion level. Combined data from 3 independent experiments. (D) Tumor cell recognition by TCR1, TCR-ESO and TCR3 transduced human PBMCs. Each 5x10⁴ effector and target cells (1:1) were co-cultured in duplicates for 16-18h. IFNγ levels were measured by ELISA. The experiment was repeated once and both yielded similar results. See also Supplementary Figure 1 .

Figure 2

Affinity comparison of NY-ESO-1_157-165 specific TCRs isolated from different source. (A) Recognition of T2 cells loaded with different amount of NY-ESO-1_157-165 peptides by different NY-ESO-1_157-165 specific T cells. Human PBMCs were transduced with NY-ESO-1_157-165 TCRs from ABab-A2 mice (TCR-ESO), human donors (Cy1, Cy3, Cy4, S09 and TCR 1G4) or in vitro mutagenized TCR from human (TCR 1G4-α95LY). Cy2 TCR was used as negative control. 10⁴ transduced CD8 were cultured with 10⁴ T2 cells loaded with different amounts of NY-ESO-1_157-165 wildtype peptide as indicated overnight. (B) IFNγ level was normalized to the highest secretion level at peptide concentration of 10^-6 M. One-site specific binding curves were then calculated from the normalized IFNγ secretion level. Pooled data from three human donors are shown. The assay was repeated in two or three independent experiments with each donor with similar results. (C) Comparisons of the EC50 of the TCRs calculated from (B). one way ANOVA was applied to compared the EC50s and * indicates a P value <0.05 and **** indicates a P value <0.0001. (D) Recognition of tumor cell lines expressing NY-ESO-1 and HLA-A*02:01. 5x10⁴ transduced enriched human CD8 T cells were cultured overnight with 5x10⁴ tumor cells. Expression of NY-ESO-1 and/or HLA-A2 is indicated. The T cell recognition was assessed by human IFNγ ELISA with the culture supernatant. The assay was repeated in two independent experiments with each human donor, three human donors were used in this experiment. See also Supplementary Figure 2 .

Figure 3

Adoptive T cell therapy of HHDxRag^-/- mice bearing mice with NY-ESO-1_157-165 specific T cells. HHDxRag^-/- mice were inoculated with 5x10⁶ MC703.gCm tumor cells and treated with 5x10⁵ transduced HHD CD8 T cells. (A) Human TCRVβ FACs staining (TCR-ESO: Vβ8; TCR 1G4-α95LY: Vβ13.1) of NY-ESO-1_157-165 TCR transduced HHD splenocytes. Gate: CD3⁺ lymphocytes. (B) Illustration of the organization of the NY-ESO-1_157-165 triple epitope cassette introduced into MC703 cells. (C) Recognition of MC703-NY cell line by NY-ESO-1_157-165 TCR transduced HHD splenocytes. 10⁴ transduced HHD splenocytes were cultured overnight with different numbers of MC703.NY cells. The experiment was done in duplicates. Non-transduced HHD splenocytes were used as negative control. (D) MC703-NY tumor growth curve in HHDxRag^-/- mice. The dashed line indicates the day of adoptive T cell transfer. The experiment was done in parallel for the two NY-ESO-1 TCRs and irrelevant T cells. Upper panel: average tumor size curve. Lower panels: tumor curves of individual mice treated with the three different T cells (TCR-ESO, TCR 1G4-α95LY, irrelevant TCR). (E) Specific CD8 T cell counts in the blood of tumor-bearing mice on day 10, 20 and 35 after T cell transfer. The experiment was repeated once (combination of both experiments, each experiment includes 4-5 mice/group).

Figure 4

TCR cross-reactivity tests. (A) Alanine scan of human CD8 T cells transduced with TCR-ESO, TCR 1G4 and TCR 1G4-α95LY. Alanine scan peptides (see Supplementary Table 1 for sequences) were loaded at indicated amount onto 10⁴ T2 cells and cultured with the same number of TCR-transduced human CD8 T cells overnight. IFNγ release was measured by ELISA. The IFNγ levels were normalized by T cell secreted IFNγ level stimulated with the wildtype peptide. The experiment was performed with three human donors, each in duplicates. A representative experiment is shown. (B) X-scan to determine the recognition motif of the NY-ESO specific TCRs. 10⁴ T2 cells were loaded with 10^-8 M NY-ESO-1_157-165 x-scan peptides, and co-cultured with 10⁴ TCR-ESO, TCR 1G4 or TCR 1G4a95 transduced huPBMCs overnight. Secreted IFNγ amounts were measured by ELISA. Upper: The recognition motif sequence Logo of the different TCRs. Lower: heatmaps indicating the changes of IFNγ level by the x-scan peptides compared to wildtype NY-ESO-1_157-165. (C) Crystal structure of the TCR1G4-MHC class I complex (pdb 2bnq) (32) with a magnified view into the peptide binding site at the right. In the crystal structure, Thr95 and Ser96 in the TCR-α chain do not directly contact the MHC α-chain (see magnification at the right, top). The magnification at the right, bottom shows a model of the TCR1G4a95-MHC class I complex, in which Thr95 and S96 are exchanged to a preferred conformer of leucine and tyrosine, respectively. Compared to Thr95 and Ser96, Leu95 and Tyr96 have longer side-chains which may form a hydrophobic cluster with the opposite Ala158, Thr168 and Tyr159 of the MHC α-chain and therefore increase the affinity of the complex. Based on the model, the bound peptide is not expected to directly contact the mutated residues.