Session: 401. Blood Transfusion: Poster III
Hematology Disease Topics & Pathways:
Research, Fundamental Science, Translational Research
We performed in silico binding analysis of Rh antigens and HLA loci using the machine learning with artificial neural networks. We used the NetMHCpan-4.1 and NetMHCIIpan-4.1 algorithms to predict HLA-peptide binding. In NetMHCpan-4.1 analysis, strong binding was scattered and did not show significant positions (hot spots). However, in NetMHCIIpan-4.1 analysis, the HLA class II pilot study showed a few clustered hotspots. Rh antigens (RHD*01, RHD*01.01, RHD*01.W.1, RHD*01.W.2, RHD*01.W.3, and RHCE*01) on HLA class II (HighQ-DRB, HighQ-DP, HighQ-DQ, DRB1, DRB3, DRB4, DRB5, DP, and DQ) because HLA class II molecules present RBC antigens via antigen-presenting cells (APCs) to interact them to T-cell receptors. We investigated the frequency and hotspots of HLA-DRB1 in four ethnic groups (Caucasian, African American, Hispanic, and Asian) and used the population frequencies of HLA alleles from the Allele Frequency Net Database.
The RhD and RhCE antigens showed several distinct hotspots for each of the HLA-DRB, -DQA-DQB, and DPA-DPB peptides of HLA class II. In in silico analysis, hot spots were in the intracellular, transmembrane, and exofacial region, depending on HLA-DRB, -DQA-DQB, and DPA-DPB. For RhD and RhCE antigens in HLA-DRB, the amino acid start position of the hot spot (strong binding site) had 15 amino acid positions. They were identified: RhD (38th, 54th, 98th, 125th, 153rd, 161st, 165th, 239th, 250th, 269th, 303rd, 347th, 358th, 390th, and 407th) and RhCE (38th, 54th, 62nd, 124th, 130th, 161st, 165th, 239th, 250th, 267th, 269th, 303rd, 358th, 390th, and 407th). The different positions unique to RhCE were 62nd, 124th, 130th, and 267th, which were not present in RhD antigens. The unique positions in RhCE antigens indicate variations that may impact the alloimmunization of RhCE antigens, compared to the RhD antigens. The amino acids in the hot spot of RhD antigens in the HLA-DRB were as follows: 38th (amino acid start position of hot spot) LEDQKGLVA (core amino acid), 54th LTVMAAIGL, 98th LSQFPSGKV, 125th ISVDAVLGK, 153rd LRMVISNIF, 161st FNTDYHMNM, 165th YHMNMMHIY, 239th FNTYYAVAV, 250th VTAISGSSL, 269th YVHSAVLAG, 303rd LISVGGAKY, 347th LVLDTVGAG, 358th MIGFQVLLS, 390th LKIWKAPHE, and 407th FWKFPHLAV. In HLA-DRB, each of the RhD antigens had the same core amino acids (epitope peptide), except for RHD*01W.1. A hot spot of RHD*01W.1 in HLA-DRB changed to p.Val270Gly (269th YGHSAVLAG), unlike a hot spot of RHD*01W.2, RHD*01W.3, and RHD*01.01. RhD and RhCE antigens showed significant differences in hot spots across the four ethnic groups. In Caucasians, HLA-DRB1*15:01 had the highest frequency, and showed a hot spot of LSQFPSGKV at the 98th amino acid start position in the RhD antigen, and a hot spot of LKIWKAPHV at the 390th amino acid start position in the RhCE antigen.
Our findings highlight the significance of immunogenicity between Rh antigens and HLA-DR, suggesting the potential clinical utility of predicting antibody development in blood transfusions. This in silico approach offers novel insights into understanding and managing alloimmunization events, particularly in patients with multiple alloantibodies when RBC transfusion is required. This study is the first attempt to use these bioinformatics methods to predict the immunogenicity of Rh antigens and HLA class II.
Disclosures: No relevant conflicts of interest to declare.