Our pathogenic pasts can’t hide

The new study used the same TCR sequencing data from the 666 individuals and increased the predictive power of that dataset by adding high-resolution HLA typing data. The authors aimed to determine if they could use the TCR sequence data to identify unknown immune exposures. As a first step, the authors investigated TCR co-occurrence patterns—TCRs that correlate with each other—among the entire study cohort (see example in Figure). The authors reasoned that a co-occurring set of TCRs may be indicative of a common pathogenic exposure. However, they realized it could also be possible for TCR co-occurrence patterns to be driven by HLA type. Indeed, their initial analyses identified two classes of co-occurring TCR pairs: one class demonstrated strong association with a shared HLA allele while the other class had a much weaker HLA allele association.

The authors used a clustering algorithm to link co-occurring TCRs together that were strongly correlated, this analysis identified 28 clusters and the number of TCRs contained within each cluster ranged anywhere from 7 to 386. From there, the authors found that nearly all of the clusters were strongly associated with at least one HLA allele. This finding prompted the authors to then analyze co-occurrence patterns within smaller subsets of individuals with specific HLA alleles. For example, in looking at the TCR associations with the A*02.01 HLA allele, they found that eight of the top ten TCRs are known to be responsive to exposure to viral antigens. In this analysis overall, they found that the more common HLA alleles are associated with greater numbers of TCRs as compared to those that are less common.

The next goal of the study was to determine if TCR clusters within HLA alleles are indicative of common immune exposure. This analysis was conducted for individual HLA alleles and included only the subset of individuals that were positive for a given HLA allele. The results were very encouraging when they found that several of the most highly significant TCR clusters contained TCRs that had previously been independently associated with various pathogens. Among the top five most significant clusters, associations with pathogens such as influenza virus, parvovirus B19, CMV, and Epstein-Barr virus were found.

This study generated a number of new insights into how to unlock the immunological history stored in our immune cells. When asked about the most significant contributions, Dr. Bradley commented, “In this paper, we introduce new tools for analysis of immune repertoires and we apply these tools to make some interesting observations about the impact of common pathogens (e.g., flu, EBV, and CMV) and immune genotype (HLA) on repertoire structure. These observations support the idea that T cell receptor (TCR) repertoire sequences carry interpretable information about our past immune exposures.”

The authors have exciting follow-up studies planned. “We want to understand which features of a person's immune exposure history and/or genotype may correlate with these TCR clusters. For some of the clusters we have hypotheses based on the literature; for others we don't have a clue,” said Dr. Bradley. New types of data will be incorporated into these future studies, which may increase the ability to identify triggers of immune response, as described by Dr. Bradley, “One plan is to generate data on antibody responses present in these samples, which could identify infectious or autoimmune exposures as correlates. We also plan to perform a genome-wide covariation analysis that could identify genetic correlates outside the HLA region.”