Quantifying how TCR sequence variation affects T cell fate at single-cell resolution

Abstract

Abstract Mucosal-Associated Invariant T (MAIT) and Natural Killer T (NKT) cells demonstrate that some T cell transcriptional fates are driven by the T cell antigen receptor (TCR). Though activation through the TCR is integral to T cell differentiation, the contribution of TCR sequence features to other T cell transcriptional fates remains yet to be defined. In this study, we identify how αβ V(D)J recombination affects T cell differentiation at single-cell resolution. By applying Canonical Correlation Analysis (CCA) to 340,557 TCR clones collected from 256 individuals, we define a set of TCR scoring functions that quantify transcriptional fate predispositions conferred by the TCR. Unsurprisingly, the strongest fate predispositions correspond to cognate peptide presentation molecules: MR1- or CD1d-restricted PLZFhigh innate-like (MAIT/NKT) transcriptional fate (72.6-fold increase in fate likelihood for top percentile compared to bottom percentile TCRs, AUC = 0.85) and MHC class I-restricted CD8 fate versus MHC class II-restricted CD4 fate (18.7-fold increase, AUC = 0.75). However, our results also uncover that hydrophobic CDR3 residues promote regulatory T cell transcriptional states in both the CD8 and CD4 lineages, and that TCR sequence features preferred by thymic positive selection continue to promote memory formation in the periphery. Applying our TCR scoring functions to 10× dextramer-labeled T cells reveals that not all antigen-specific TCRs are equally capable of mounting an effector response. Even among T cells that recognize the same antigen, biophysical variation in the TCR sequence directs which T cells undergo transcriptional reprogramming to form immunological memory. Supported by grant from NIH (T32GM007753)