Potentialities of Bioinformaticaly Predicted Linear B-cell Epitopes on Dengue prM Protein

Abstract

Predicting linear B-cell epitopes using bioinformatic tools is attractive as an initial step for screening such epitopes with potential use in Dengue (DENV) diagnosis and therapeutics, for their high speed and low cost. B-cell epitopes on the premembrane protein (prM) of DENV are major targets of inducing humoral immunity. However, B-cell epitopes on prM protein of DENV have not been well characterized. In this study, three tools namely: BepiPred, Ellipro and SVMTrip, were used to predict linear B cell epitopes of dengue prM protein and results were compared. Further, a concise verification against available biochemical-assay positive data was carried out. The prM protein sequences of 50 strains of each of the four DENV serotype with temporal differences and geographical variations were analysed. Predictions yielded 12 epitopes with variable lengths (5– 50 aa). Most predicted epitopes, at least partially, overlap with regions shown to generate antibodies or are recognized by natural antibodies in other studies. The predicted epitopes of the three tools further demonstrated good agreement, where the peptide location had been predicted as epitopes by more than one tool. Collectively, results demonstrate the potentiality of computer based tools in predicting truly immunogenic epitopes. In terms of conservancy of predicted epitopes, many showed low conservancy levels among the four serotypes (less than 70%). Such epitopes could contribute to antibody dependent enhancement of secondary dengue infections rather than neutralization as has been documented. Overall, serotype specific conservation seemed to be higher in epitopes from DENV3 and DENV4, but not in the other two serotypes. Out of the predicted epitopes only two epitopes, EP9 and EP10, have conservation higher than 70%, indicating a potential use of them as a universal vaccine candidate. In conclusion, based on the predictions observed in the current study, the bioinformatic approach is found to be a good and positive initial step to screen potential linear epitopes in proteins.