Malaria caused by Plasmodium knowlesi species has become a public health concern, especially in Malaysia. Plasmodium knowlesi parasite which originates from the macaque species, infects human through the bite of the Anopheles mosquitoes. Research on malaria vaccine has been a continuous effort to eradicate the malaria infection, yet there is no vaccine against P. knowlesi malaria to date. Apical membrane antigen 1 (AMA1) is a unique surface protein of all apicomplexan parasites that plays a crucial role in parasite-host cell invasion and thus has been a long-standing malaria vaccine candidate. The selection of protective epitopes in silico has led to significant advances in the design of the vaccine. The present study aimed to employ bioinformatics tools to predict the potential immunogenic B- and T-cell epitopes in designing malaria vaccine targeting P. knowlesi AMA1 (PkAMA1). B-cell epitopes were predicted using four bioinformatics tools, i.e., BepiPred, ABCpred, BcePred, and IEDB servers whereas T-cell epitopes were predicted using two bioinformatics servers, i.e., NetMHCpan4.1 and NetMHCIIpan-4.0 targeting human major histocompatibility complex (MHC) class I and class II molecules, respectively. The antigenicity of the selected epitopes computed by both B- and T-cell predictors were further analyzed using the VaxiJen server. The results demonstrated that PkAMA1 protein encompasses multi antigenic regions that have the potential for the development of multi-epitope vaccine. Two B- and T-cell epitopes consensus regions, i.e., NSGIRIDLGEDAEVGNSKYRIPAGKCP (codons 28-54) and KTHAASFVIAEDQNTSY RHPAVYDEKNKT (codons 122-150) at domain I (DI) of PkAMA1 were reported. Advancement of bioinformatics in characterization of the target protein may facilitate vaccine development especially in vaccine design which is costly and cumbersome process. Thus, comprehensive B-cell and T-cell epitope prediction of PkAMA1 offers a promising pipeline for the development and design of multi-epitope vaccine against P. knowlesi.

Plasmodium knowlesi is the most common zoonotic parasite associated with human malaria infection in Malaysia. Apical membrane antigen 1 (AMA1) protein in the parasite plays a critical role in parasite invasion into host cells. To date, there is no complete three-dimensional ectodomain structure of P. knowlesi AMA1 (PkAMA1) protein. The knowledge of a protein structure is important to understand the protein molecular functions. Three in silico servers with respective structure prediction methods were used in this study, i.e., SWISS-MODEL for homology modeling and Phyre2 for protein threading, which are template-based modeling, while I-TASSER for template-free ab initio modeling. Two query sequences were used in the study, i.e., native ectodomain of PkAMA1 strain H protein designated as PkAMA1-H and a modified PkAMA1 (mPkAMA1) protein sequence in adaptation for Pichia pastoris expression. The quality of each model was assessed by ProSA-web, QMEAN and SAVES v6.0 (ERRAT, Verify3D and Ramachandran plot) servers. Generated models were then superimposed with two models of Plasmodium AMA1 deposited in Protein Data Bank (PDB), i.e., PkAMA1 (4UV6.B) and Plasmodium vivax AMA1 (PvAMA1, 1W81) protein structures for similarity assessment, quantified by root-meansquare deviation (RMSD) value. SWISS-MODEL, Phyre2 and I-TASSER server generated two, one and five models, respectively. All models are of good quality according to ProSA-web assessment. Based on the average values of model quality assessment and superimposition, the models that recorded highest values for most parameters were selected as best predicted models, i.e., model 2 for both PkAMA1-H and mPkAMA1 from SWISS-MODEL as well as model 1 of PkAMA1-H and model 3 of mPkAMA1 from I-TASSER. Template-based method is useful if known template is available, but template-free method is more suitable if there is no known available template. Generated models can be used as guidance in further protein study that requires protein structural data, i.e., protein-protein interaction study.