Predicting peanut allergic reaction with integrative single cell models of T cell immuno-synapse
Xintong Chen1, David Chiang2, Cecilia Berin2, Bojan Losic1
1Icahn Institute for Genomics & Multiscale Biology, Mount Sinai School of Medicine, Department of Genetics, NYC; 2Mount Sinai School of Medicine, NYC
A dynamical, molecular-level understanding of immune priming and clonal expansion in the context of peanut allergy is now possible with single cell RNA-seq (scRNA-seq) technology. In this work we informatically characterize the T cell receptor – (peanut allergen – MHC) interaction in peanut-responsive CD4+ T cells in peripheral blood via scRNA-seq to construct a model of the receptor-epitope synapse which predicts biomarkers for clonal expansion and allergic reaction.
scRNA-seq was generated for 212 peanut-responsive CD154+ cells from 5 peanut allergy (PA) subjects and 122 antiCD3/CD28-activated CD154+ cells from 3 PA subjects. We also collected Tregs from freshly isolated PBMCs from PA or HC subjects (97 resting Tregs) as an additional reference population, for a total of 431 cells. We reconstructed the full complementarity determining region 3 (CDR3) directly from the sc-RNA-seq data to characterize both (alpha and beta) chains of the cells. The reconstructed recombinant sequences contained nearly the full expected length of the TCR VDJ region and we detected at least one productive alpha chain in 68.6% of the cells, a productive beta chain in 74.8% of the cells, and productive alpha beta pairs in 59.7%. Clonal expansion was clearly observed only in peanut responsive cells.
By aligning the assembled CDR3 sequence in our data against previously published allergen (Ara h1) binding hotspots and then combining with a clonal expansion gene signature obtained via differential expression and causal network inference, we were able to create features for a machine learning approach that accurately predicts allergic response using just scRNA-seq data from peripheral blood.