Cross-reactivity of Influenza and ADAMTS13 Peptides in Acquired TTP

Abstract

Thrombotic thrombocytopenic purpura (TTP) is a rare but deadly hematologic disease with a prevalence of ~10 cases per million people per year and has a ~90% mortality rate. TTP is caused by low activity of metalloprotease called ADAMTS13, which is responsible for cleaving von Willebrand Factor (vWF), a multimeric glycoprotein that is involved in blood coagulation. Low activity of ADAMTS13 keeps vWF in its multimer state, triggering blood clotting more frequently. In acquired TTP, ADAMTS13 is inhibited by auto-antibodies, but the underlying mechanism of the autoimmunity is still unknown. TTP onsets are frequently reported after infection or vaccination of influenza A virus, especially H1N1 strain. Recent studies suggest that influenza may trigger TTP via molecular mimicry, a process where autoreactive T cells recognize influenza peptides that are cross-reactive with ADAMTS13 peptide. HLA DRB1*11 repeatedly reported as predisposing factor of TTP, while HLA DRB1*04 is considered as a protective factor. Two ADAMTS13 CUB2 domain peptide, FINVAPHAR (position 1324-1332) and LIRDTHSLR (position 1355-1363) were found to bind HLA DRB1*11 and HLA DRB1*03 respectively and can trigger CD4+ T cells response. This research aims to model molecular mimicry between influenza and ADAMTS13 peptide and to predict which influenza peptide can cross-react with ADAMTS13 peptides. We use netMHCIIpan to predict MHC binders and build BLAST-like model to score T cell cross-reactivity between influenza and ADAMTS13 peptide. PS-SCL data on T cell response is incorporated to improve the T cell reactivity prediction. Our model is able to bring up the previously known MHC-binders from ADAMTS13, LIRDTHSLR, as one of the most cross-reactive peptides. It aligns with influenza C peptide VLIADAKGL (nucleoprotein pos. 189-197). We also found several other potentially cross-reactive ADAMTS13 peptides that align with peptides from influenza A and SARS-CoV-2. Furthermore, we applied this model to another environmentally-triggered autoimmune disease, multiple sclerosis. Our model predicts a self-foreign peptide pair that is proven to be cross-reactive in experiments as the most cross-reactive peptide pair. Taken together, our model is able to model molecular mimicry in autoimmune disease and predict the cross-reactive