What can I TIL you? Decoding TCR antigens

See allHide authors and affiliations

Science Immunology  02 Feb 2018:
Vol. 3, Issue 20, eaat0810
DOI: 10.1126/sciimmunol.aat0810


Novel unbiased strategy for identification of peptide antigens bound by tumor-infiltrating lymphocyte T cell receptors.

Over the last few years T cells have been tapped as therapeutic interventions against end stage cancer, either by creation of bespoke T cells that attack key antigens of target cells (i.e., chimeric antigen receptor T cells) or predicting and exploiting neoantigens for therapeutic vaccines. However, in many situations—including infection, autoimmunity, and oncology—we need to more thoroughly define the antigens bound by a given T cell receptor (TCR) to identify novel therapeutic targets. Here, Gee et al. focused on tumor-infiltrating lymphocytes (TILs) and applied an unbiased strategy to assess what an assayed TCR recognizes. They developed a yeast-display library wherein each yeast displays one of millions of uniquely tagged peptide–human lymphocyte antigen (pHLA) complexes. Using beads coated with a single TCR specificity, multiple rounds of selection within the pHLA yeast-display library lead to convergence upon a subset of pHLA. Enriched peptides are used to optimize TCR antigen prediction using multiple algorithms. Gee et al. collected T cells from two male patients with colorectal adenocarcinoma for antigen discovery. Beginning with hundreds of CD8 T cells (from tumor and healthy tissue), they selected 20 candidate TCRs, based on effector function and expansion within the tumor. Four TCRs yielded enrichment of target peptides from the pHLA yeast-display library and thus allowed antigen prediction. Two TCRs recognized a set of similar peptides and were predicted to recognize U2AF2 (overexpressed in multiple tumor cell lines and able to activate the T cells in vitro). The other two TCRs were each restricted to a single patient. Gee et al. performed whole exome sequencing of the tumors and predicted neoantigens using multiple algorithms. Both these antigens and the human proteome were used as targets in the prediction algorithms. Notably, the majority of TCRs were predicted likely to recognize non-mutated self-antigens. This strategy will permit discovery of novel immunogenic peptide targets and could be extended to infectious disease and autoimmunity. It will be important to clarify the key factors limiting efficacy of this strategy (here successful with 4/20 TCRs), including HLA type, TCR peptide affinity, half-life of peptide binding to HLA or a combination of this and other technical and immunological factors.

Highlighted Article

View Abstract

Navigate This Article