P01-04
Cross-reactivity of T cell receptors against HCoV through three-dimensional structure prediction
Ao KIKUCHI *1, Toru EKIMOTO1, Tsutomu YAMANE2, Shuntaro CHIBA2, Kanako SHIMIZU3, Shin-ichiro FUJII3, Mitsunori IKEGUCHI1, 2
1Graduate School of Medical Life Science, Yokohama City University,
2Center for Computational Science, RIKEN
3Center for Integrative Medical Science, RIKEN
( * E-mail: w235417a@yokohama-cu.ac.jp )
In cellular immunity, killer T cells play a major role in eliminating virus-infected cells and preventing the spread and severity of infection. One of the crucial steps in triggering the immune response is the recognition of the viral antigen peptide-human leukocyte antigen (HLA) complex (pHLA) presented on the surface of infected cells by the T cell receptor (TCR) of the killer T cell via the formation of a complex of pHLA and TCR. Most TCRs have antigen specificity, but there are also cross-reactive TCRs that recognize multiple types of pHLA. Recently, several antigen peptides derived from human coronavirus (HCoV) with high affinity to HLA-A*24:02, the most common HLA type in Japanese, were identified, and it was revealed that there are killer T cells that cross-react with these peptides (Shimizu, K. et al. Commun Biol. 2021). This suggests that the killer T cells that worked with seasonal coronaviruses may also work when infected with novel coronaviruses, and therefore prediction of TCR cross-reactivity is expected to lead to elucidation of cross-reaction mechanisms and development of therapeutic and preventive measures to enhance immune responses. However, there is no method to predict TCR cross-reactivity with high accuracy. In this study, we aimed to develop a method to classify TCRs that react to HCoV-derived peptide-HLA-A*24:02 complex (pHLA) based on their three-dimensional structures. From the pHLA and TCR sequences, we predicted a structure of the pHLA-TCR complex using TCRmodel2, an AlphaFold2-based structure prediction tool (Yin, R. et al. Nucleic Acids Res. 2023) and constructed a method for classifying the reactivity of the TCR using the confidence score of the binding interface estimated from the predicted structure (Jumper J et al. Nature. 2021) to classify TCR reactivity. As a result, we succeeded in classifying about 90% of 28 types of TCR-pHLA reactivity including cross-reactivity.
