Identification of the peptides recognized by individual T cells is important for understanding and treating immune-related diseases. Current cytometry-based approaches are limited to the simultaneous screening of 10-100 distinct T-cell specificities in one sample. Here we use peptide-major histocompatibility complex (MHC) multimers labeled with individual DNA barcodes to screen >1,000 peptide specificities in a single sample, and detect low-frequency CD8 T cells specific for virus- or cancer-restricted antigens. When analyzing T-cell recognition of shared melanoma antigens before and after adoptive cell therapy in melanoma patients, we observe a greater number of melanoma-specific T-cell populations compared with cytometry-based approaches. Furthermore, we detect neoepitope-specific T cells in tumor-infiltrating lymphocytes and peripheral blood from patients with non-small cell lung cancer. Barcode-labeled pMHC multimers enable the combination of functional T-cell analysis with large-scale epitope recognition profiling for the characterization of T-cell recognition in various diseases, including in small clinical samples.

Prediction of T-cell receptor (TCR) interactions with MHC-peptide complexes remains highly challenging. This challenge is primarily due to three dominant factors: data accuracy, data scarceness, and problem complexity. Here, we showcase that “shallow” convolutional neural network (CNN) architectures are adequate to deal with the problem complexity imposed by the length variations of TCRs. We demonstrate that current public bulk CDR3β-pMHC binding data overall is of low quality and that the development of accurate prediction models is contingent on paired α/β TCR sequence data corresponding to at least 150 distinct pairs for each investigated pMHC. In comparison, models trained on CDR3α or CDR3β data alone demonstrated a variable and pMHC specific relative performance drop. Together these findings support that T-cell specificity is predictable given the availability of accurate and sufficient paired TCR sequence data. NetTCR-2.0 is publicly available at https://services.healthtech.dtu.dk/service.php?NetTCR-2.0.

Predicting epitopes recognized by cytotoxic T cells has been a long standing challenge within the field of immuno-and bioinformatics. While reliable predictions of peptide binding are available for most Major Histocompatibility Complex class I (MHCI) alleles, prediction models of T cell receptor (TCR) interactions with MHC class I-peptide complexes remain poor due to the limited amount of available training data. Recent next generation sequencing projects have however generated a considerable amount of data relating TCR sequences with their cognate HLA-peptide complex target. Here, we utilize such data to train a sequence-based predictor of the interaction between TCRs and peptides presented by the most common human MHCI allele, HLA-A*02:01. Our model is based on convolutional neural networks, which are especially designed to meet the challenges posed by the large length variations of TCRs. We show that such a sequence-based model allows for the identification of TCRs binding a given cognate peptide-MHC target out of a large pool of non-binding TCRs.

The promiscuous nature of T-cell receptors (TCRs) is fundamental for our ability to recognize a large range of pathogens; however, this feature makes it challenging to understand and control Tcell recognition 1. Existing technologies provide limited information about the key requirements for T-cell recognition and the ability of TCRs to cross-recognize structurally related elements 2,3. Herein we present a proof-of-concept of a novel 'one-pot' strategy to establish the patterns that govern TCR recognition of peptide-major histocompatibility complex (pMHC). We determine the affinity-based hierarchy of TCR interactions with MHC loaded with peptide variants, and apply this knowledge to understand the recognition motif, here termed the TCR fingerprint. The TCR fingerprints of 16 different TCRs were identified and used to predict and validate crossrecognized peptides from the human proteome. The identified fingerprints differed amongst TCRs recognizing the same epitope, demonstrating the value of this strategy for understanding T-cell interactions and assessing potential cross-recognition prior to selection of TCRs for clinical development.

Narcolepsy Type 1 (NT1) is a neurological sleep disorder, characterized by the loss of hypocretin/orexin signaling in the brain. Genetic, epidemiological and experimental data support the hypothesis that NT1 is a T-cell-mediated autoimmune disease targeting the hypocretin producing neurons. While autoreactive CD4+ T cells have been detected in patients, CD8+ T cells have only been examined to a minor extent. Here we detect CD8+ T cells specific toward narcolepsy-relevant peptides presented primarily by NT1-associated HLA types in the blood of 20 patients with NT1 as well as in 52 healthy controls, using peptide-MHC-I multimers labeled with DNA barcodes. In healthy controls carrying the disease-predisposing HLA-DQB1*06:02 allele, the frequency of autoreactive CD8+ T cells was lower as compared with both NT1 patients and HLA-DQB1*06:02-negative healthy individuals. These findings suggest that a certain level of CD8+ T-cell reactivity combined with HLA-DQB1*06:02 expression is important for NT1 development.