Building an atlas to predict alloimmunity


Transplantation is often a life-saving therapy, whether it is bone marrow for a patient bearing a malignancy, or a solid organ for a patient with organ failure. In either setting, however, differences between the donor and recipient proteomes can be detected by T cells and lead to alloimmunity, with the potential to completely negate the therapeutic effect. A major pathway of alloimmunity – ‘indirect recognition’ – involves reactivity directed to peptide sequences that differ between donor and recipient. Such variants are encoded across the human genome, starting with the hypervariable Major Histocompatibility Complex (MHC), which also encodes the proteins through which these genomic differences become visible to T cells.

Current clinical approaches to donor-recipient matching are generally limited in several respects: (1) they usually consider only MHC (and rarely other variable loci), and (2) even for the variants that are considered, matching algorithms do not consider the potential for mismatches to form MHC-binding peptides, which is a gating event for indirect T cell recognition. In this project, we aim to address these limitations using a highly-multiplexed and programmable peptide-MHC binding assay. With this tool, we are working to generate a high-dimensional atlas that maps the binding of common human genetic variants to common human MHC proteins. We hypothesize that this atlas can predict the likelihood of alloimmunity in any given donor-recipient pair.