Naïve adaptive immune receptor repertoires in celiac disease assessed by machine learning; impact of the HLA-DQ2.5 allotype on the TCR repertoire

Abstract

The adaptive immune receptor repertoire (AIRR) - the collection of an individual’s B-cell and T-cell receptors (BCRs and TCRs, respectively) - encodes cumulative immune history and is shaped by both germline genetics and environmental exposures. Skewed repertoires have been linked to infections, vaccination responses and autoimmune diseases such as celiac disease (CeD) where biased usage of immunoglobulin and T-cell receptor genes reactive to disease relevant antigens has been reported. Motivated by evidence that germline variation, notably human leukocyte antigen (HLA), influences naïve AIRRs and by prior machine-learning studies that classified CeD using naïve BCR AIRR-sequencing (AIRR-seq) data, we applied machine learning analysis to naïve CD4+ TCR and naïve BCR AIRR-seq repertoires to test whether repertoire features can distinguish subjects with CeD from controls and to identify drivers of such classification. Naïve CD4+ TCR repertoires yielded moderate diagnosis classification, but this signal was largely explained by enrichment of the HLA-DQ2.5 allotype. TCR variable gene frequencies predicted HLA-DQ2.5 status with high accuracy and controlling for HLA-DQ2.5 abolished TCR-based diagnosis classification. In contrast, naïve BCR repertoires could not be used to classify CeD. Overall, our findings show that germline HLA variation significantly affects naïve TCR composition and thereby indirectly facilitates moderately successful CeD status classification.