COMBINATORIAL ALGORITHMS PROVIDE FINE MAPPING OF EPITOPES RECOGINED BY HLA CLASS I MONOCLONAL ANTIBODIES.

3.1
#68
COMBINATORIAL ALGORITHMS PROVIDE FINE MAPPING OF EPITOPES RECOGINED BY HLA CLASS I MONOCLONAL ANTIBODIES.
Martin Maiers , Geziena M.T. Achreuder , Arend Mulder , Jar Haw Lee , Martin Gutierrez , Carol Machan , Hugo A. Araujo , Edmond J. Yunis and Marcelo A. Fernandez-Vina . Minneapolis MI, National Marrow Donor Program, Information Systems ; Leiden Netherlands, Leiden University Medical center ; Canoga Park CA, One Lambda, Inc. ; Washington DC, Georgetown University, Department of Oncology and CW Bill Young DoD Marrow Donor Program and Boston MA, Dana Farber Cancer Institute .

Amino acids located in both alpha helices and connecting loops of the HLA class I molecules appear to determine the antibody reactivity. Core residues for most allo-epitopes or the contribution of non-exposed amino acids have not been entirely elucidated. We developed computational approaches to correlate antibody reactivity with amino acids on the HLA molecules. We analyzed 710 cells studied by the Serology Component of the 13^th IHWS; typed for HLA-A and B loci at allele level and tested with 274 mAb by cytotoxicity. The specificity of the mAb was determined by correlation/tail analysis and by two alternative computational approaches associating mAb reactivity with amino acids at the alpha-1 and 2 domains (putative epitopes, pEp) .The first algorithm examined the 8 individual amino acids with he highest correlation of all possible residue combinations (40320). Another algorithm examined the 20 top residues, paired them and selected the top duplet(s). Further iterations of residue aggregation was performed until exceeding correlations were not found. For 268 mAb, either of these algorithms or both had higher correlations than those obtained by formal serologic analysis. This trend was not observed in 6 reagents associated with a few or low frequency alleles. Clustering the sequences of allleles with differential serum inclusion were developped to refine the mapping of pEp. These computational analyses do not make assumptions about number or location of residues. A refined definition and prediction of HLA serologic epitopes could improve both, donor search and matching algorithms or strategies.