Immune signatures of common exposures through co-occurrence of T-cell receptors in tens of thousands of donors

Background Memory T cells are records of clonal expansion from prior immune exposures such as infections, vaccines and chronic diseases. Some of the receptors of these expanded T cell clones in a typical immune repertoire are highly public (present in many individuals) because they respond to the same peptide from a prevalent immune exposure, presented by the same Human Leukocyte Antigen (HLA) allele. Only a tiny fraction of public T-cell receptor β sequences (TCRs) have known associations with exposures or specific peptides. Methods We mined the TCR repertoires of tens of thousands of donors to define “ECOclusters” (Exposure Co-Occurrence clusters): clusters of public TCRs that tend to occur in the same donors. First, we built models to infer donor HLA type from the TCR repertoire, then associated public TCRs with HLA alleles. Next, we derived co-occurrence clusters of TCRs responding to antigens presented by the same HLA allele, then combined those clusters by co-occurrence across HLA alleles. Each such cross-HLA ECOcluster putatively represents a public TCR signature of a single exposure. Results We constructed sensitive, specific models to predict the presence of 220 HLA alleles from TCR repertoires and clustered 8,618,285 HLA allele-associated TCRs to define 11,058 ECOclusters. Using serologically labeled repertoires, we identified ECOclusters associated with HSV-1, HSV-2, EBV, Parvovirus, Toxoplasma gondii , Cytomegalovirus (CMV) and SARS-CoV-2, and constructed sensitive, specific classifiers of exposure. We explore the 50,217 TCRs in the CMV ECOcluster in detail, including diagnostic performance of each HLA-associated component and intersection with public TCR databases. ECOclusters represent a step toward deciphering the ledger of immune exposure history encoded by the T-cell repertoire.