We primarily focused our effort in the structural understanding of T cell development and T cell mediated immunotherapy. T cells are developed in thymus to become MHC-restricted. While it is generally accepted that MHC binding affinity drives T cell maturation. However, it is not clear what structural constrains imposed by MHC to shape a developing T cell receptor repertoire. In collaboration with Dr. Alfred Singer's group at NCI, we investigated the determinants for T cell receptor specificity against MHC ligands. The central feature of T cell biology is that T cells selected from the thymus utilize their antigen receptors (TCR) to specifically recognize antigenic peptides presented by the major histocompatibility complex (MHC), a characteristic referred to as MHC-restriction. However, the mechanisms leading to MHC-restriction of T cells are not fully understood. Previous structural studies of TCR and peptide/MHC complexes suggested that MHC-restriction is intrinsic to TCR structures as a result of germline-encoded CDR1 and CDR2 loops that have evolved to specifically promote contacts with MHC. In contrast, emerging evidence has shown that MHC-restriction is imposed by thymic selection in that coreceptor-independent TCR signaling in the thymus permits selection of TCRs that recognize MHC-independent ligands. Further deep sequencing analysis revealed that the overall germline V gene usage was similar in the peripheral TCR repertoire of both wild type B6 and Quad_KO mice. Nevertheless, individual TCR clones were selected at remarkably different frequencies in the presence or absence of MHC, further demonstrating the ligand specificity of TCRs was imposed by the thymic selection. The comparisons of TCR sequences and their frequencies between MHC-restricted and independent repertoires enabled us to delineate any amino acid sequence preferences both in germline-encoded CDR1 and 2, and in non-germline derived CDR3 regions specific for MHC-recognition. In an attempt to characterize the usage of both germline and non-germline regions of TCR in response to MHC, we carried out RNA-seq based deep sequencing on TCR chains from both MHC-independent and MHC-restricted animals. To address the impact of thymic selection to TCR repertoires, we also sequenced TCR repertoire from pre-selection double negative (DN) B6 thymocytes. These comparative repertoire analyses of polyclonal TCRs from different mouse strains revealed that MHC-restricted but not independent repertoires share greater number of public TCR sequences. They selectively enrich CDR3 sequences containing smaller, hydrogen-bonding amino acids but disfavor larger, hydrophobic amino acids and exclude cysteine residues in their MHC-binding site. In contrast, MHC-independent TCRs exhibiting antibody-like CDR3 amino acid compositions. The selective preference of residue composition in CDR3 but not in germline-encoded CDR1 and 2 demonstrates that MHC specificity is not intrinsic to germline-encoded TCR sequences but results from ligand-specific selections. In addition to comparative repertoire analyses, we also report the first crystal structures of MHC-independent TCRs (A11 and B12A),both recognized mPVR as their activation and selection ligand. B12A and A11 closely resemble the conventional MHC-restricted TCRs. The structural comparison of these MHC-independent TCR with those of MHC-restricted ones illustrated the lack of structural changes in CDR1 and 2, demonstrating the structures of germline V genes could not pre-determine TCR ligand specificity. Moreover, B12A and A11 TCRs recognized different epitopes on mPVR, reminiscent of antibody-antigen recognition. In summary, our results suggest that within the pre-selected repertoire TCRs are capable of recognizing a huge diversity of ligand structures and highlight the role of thymic selection in determining the TCR specificities. The comparison of TCR amino acid sequences of various pre-selection repertoires with those of mature repertoires from multiple MHC-specific strains as well as MHC-independent (MHCi) animals showed that MHC-specific thymic selection affected only non-germline encoded CDR3, restricting both their length and usage of specific amino acids, but did not affect TCR V-gene usage nor V-J pairing. Violation of these constraints may result in T cells to fail positive selection from unfavorable interactions with MHC or undergo clonal deletion due to self-reactivity. Adoptive T cell immunotherapy (ACT) treats cancer by the injection of large numbers of tumor specific T cells. While promising, ACT is not 100% effective and understanding successful cases will facilitate the improvement of ACT and other immunotherapies. Recently, ACT was successfully used to successfully treat a patient with metastatic colorectal cancer where the tumor carried the common oncogenic G12D mutation in KRAS. The success of this therapy is dependent on T cell recognition of peptides derived from mutant KRAS G12D presented by MHC class I molecules on the tumor cell surface. T cells specific for this mutation were identified to be restricted by the MHC class I molecule HLA-C*08:02 and recognized two peptide epitopes derived from mutant KRAS G12D. The two peptides were a 9mer (GADGVGKSA) and a 10mer (GADGVGKSAL), differing only in their length with the mutation at peptide position 3. Using x-ray crystallography, we solved the high resolution crystal structures of HLA-C*08:02 presenting the 9mer and 10mer peptides alone and in complex with their cognate T cell receptors (TCRs). The TCR free HLA-C structures demonstrated a bulge in the 10mer peptide compared to the 9mer, peaking at K7. In the TCR:HLA-C-10mer structure this bulge in the peptide shifted from K7 to V5 and is accommodated through hydrophobic contacts with the CDR3 chain, while the main peptide contact is through a Y97 in the CDR3, which interacts with the carboxy of peptide G4. This conformation dependent recognition of the 10mer provides a structural explanation for why the 10mer specific TCR is unable to recognize the highly similar 9mer peptide. The TCR:HLA-C-9mer structure revealed that the 9mer peptide was recognized primarily through a E51 and Y50 of the CDR2, while the CDR3 did not contribute to peptide recognition, and interacted with the HLA-C heavy chain. The CDR3 chain through Q98 contacted both the peptide at position D3 and HLA-C at R156. Through modelling, some of these contacts are likely to be maintained explaining the weak cross-reactive T cell recognition of the 10mer peptide by the 9mer specific TCR. Additionally, our structures demonstrated that both epitopes are similarly presented by HLA-C*08:02 through a direct contribution of the G12D mutation, which forms a salt bridge between peptide D3 and R156 of HLA-C*08:02. This interaction is essential for antigen presentation of the G12D KRAS peptides, suggesting the wild type KRAS peptides cannot be presented by HLA-C*08:02, explaining why they are not recognized by these TCRs. Collectively, our data provide a structural basis for effective ACT to mutant KRAS G12D and offer a blueprint for rational design of peptide vaccines and TCR based immunotherapies. In addition, by identifying the key TCR residues necessary for recognition of mutant KRAS G12D, we facilitate the tracking and identification of KRAS G12D specific TCRs in other patients carrying HLA-C*08:02 and tumors bearing mutant KRAS G12D.
Showing the most recent 10 out of 16 publications
