The NTS-DBL1α domain of the pRBC surface expressed PfEMP1 molecule has been shown to be of central importance for the virulence of the parasite. This domain is involved in rosetting [15, 19, 24, 30] and has recently been shown to induce cross-reactive antibodies able to react with different rosetting-associated variants of the molecule , therefore being a promising candidate in the development of a vaccine against severe malaria. To efficiently design and test a vaccine it is important to understand if different NTS-DBL1α variants of the DBL1α1 subtype can induce consistent antibody responses in different animals. In this study, the response to immunization with distinct NTS-DBL1α domains in different animals was analysed employing ELISA assays, live cell surface reactivity, epitope prediction and peptide array.
PfEMP1-domains expressed by parasite strains with a rosetting phenotype [19, 24, 25] were chosen for expression of recombinant protein and animal immunizations. NTS-DBL1α constructs of PAvarO and IT4var9 covered aa 1–393 and refolded from inclusion bodies, for IT4var60 a longer construct was designed (aa 1–481), which was soluble in E. coli. All proteins were monomeric and folded as judged by mobility on non-reducing gel and size exclusion profile (Figure 1). There were no differences detected as both antibody titres and functionality was similar for the antibodies generated against the different constructs (Figures 2 and 3) suggesting that monomeric state and folding are sufficient requirements for potent induction of biologically active antibodies. Further, epitopes elicited by immunizations with different constructs were similar.
Previous studies showed induction of functional antibodies towards NTS-DBL1α domains in mice and rabbits [24, 46, 56] but no attempts has been made to compare induction of antibodies in different animals regarding the epitopes inducing or targeted by them.
All immunogens in the different animals, both species and individuals, elicited a similar response when analysing titres towards homologous proteins (Figure 2). This is in contrast to what was previously reported for DBL domains of the pregnancy malaria vaccine candidate VAR2CSA where substantial differences were detected when immunizing mice, rats and rabbits . This could be possibly explained by similar induction of B-cell epitopes in different species, as well as conservation of immunogenic features in the proteins studied herein.
When comparing predicted and recognized epitopes substantial differences were found (Additional file 3). The prediction method used for this analysis identified mainly epitopes localized in the loop region of the molecule also predicted to be surface exposed. However, analysis of peptides that were in fact recognized revealed that also α-helical structures are predominant sites of immune responses. The conserved and non-exposed LARSFADIG sequence, present in SD2, was not predicted as highly immunogenic but antibody responses were frequent and high in all animals, suggesting also that conserved parts of the molecule are processed and presented on Major Histocompatibility complex (MHC) molecules. The observations presented in this study suggest that B-cell epitope prediction is informative; however, there might be a tendency for skewing towards surface epitopes that are present in unstructured loops. The analysis on epitopes recognized by immunized animals suggests that it is more common to have epitopes that span structured regions such as α-helices and that not necessarily are surface exposed. The latter could be a mechanism of immune evasion by which parasites direct the immune response towards epitopes that are not displayed on the cell surface and therefore impede the labelling of the pRBC with antibodies.
Analysis of the peptide array data for epitope recognition visualized substantial variation in between different animals, despite their isogenicity. No clear consensus was obtained when analysing different animal species immunized with the same protein (Figure 5). In addition, a limitation of this method is the fact that only linear epitopes can be detected; possibly, conformational epitopes are predominant and account for equal potency and efficiency of antibodies in different species. However, for both IT4var60 and IT4var9 there was a tendency in different species to recognize different epitopes in the N terminal part of the protein while more consensus was present concerning epitopes in SD3 of DBL1α.
In this study, surface labelling of heterologous pRBC with reagents against NTS-DBL1α-domains, did not reveal extensive cross-reactivity in heterologous parasite strains. Cross-reactivity in ELISA appears much more common for all three proteins analysed in the study. It has been suggested that coating of antigens to plastic surfaces as applied in ELISA-assays might unveil otherwise hidden epitopes [49, 56]. When analysing epitopes cross-recognized by goat IgG the conserved motif LARSFADIG is consistently present. In addition, a peptide in the end of h7 is cross-recognized by some goat IgG. This part of the molecule has also previously been indicated as possible site for generation of ELISA cross-reactive antibodies . Cross-reactivity in ELISA strongly correlates with the array cross-reactivity suggesting that most of it is due to those epitopes. These results should be taken into consideration when analysing cross-reactivity or sero-prevalence relying solely on ELISA data: despite the fact that the recombinant domain is correctly folded it might expose highly immunogenic epitopes that are not available in the full length PfEMP1 presented on the erythrocyte surface. Complementing ELISA with surface reactivity-data could minimize false positive results.